Mentors

About

Lucas Ospina is a first-year Ph.D. student in Chemical and Biomolecular Engineering (ChBE) working with Prof. Natsumi Komatsu on developing neurochemical sensors for real-time visualization of chemical signaling in the brain. He earned his B.S. in Chemical Engineering with a concentration in Nanotechnology from Massachusetts Institiute of Technology (MIT) in 2026, and has worked as a high-school STEM educator in the Republic of South Korea (Seoul and Jeju). He especially enjoys interdisciplinary research and a hands-on approach to learning. He belives the best form of mentorship occurs when the mentor-mentee work as a team and learn together.

Description of possible projects

Project Title

Neurochemical sensor development

Project Description

The student will contribute to a project at the intersection of material science, chemical engineering, and neuroscience. Given the interdisciplinary nature of the work there will be a strong emphasis on understanding how the different engineering fields come together in this project. Students will learn how DNA can be for molecular recognition and transform the optical properties of semiconducting materials. There are a couple directions for students to choose from; (1) Students can learn about the physics of the field and learn experimental techniques that were developed from those first-principles. From this the student can identify interesting functional groups that would separate certain neurochemicals from others, and propose a selective sensor design. (2) Students can learn about the physics of the field and contribute to a review paper where we discuss future directions of the field with an emphasis on first-principles rational design. Student will be listed as a co-author on this paper. (3) For students more interest in the neuroscience aspect of the project we can construct a library of interesting neurochemicals that play important roles in a neurological process. Then the student can screen that library against a batch of pre-existing sensors (high throughput screening) to find a good sensor or sensor pair that gives insight into the neurological process. For example: Being able to detect dopamine (neurochemical) spikes during rewarding experiences (stimulus).

Students will takeaway with them an intro to the fascinating nanoscale world and exposure to highly interdisciplinary research.

Major and Skill Requirements

Our research is very interdisciplinary! We welcome students from Bioengineering, Chemical Engineering, Material Science Engineering, Electrical Engineering, Mechanical Engineering, Physics, Neuroscience / Neural Engineering.

Desired skills and time commitment

• Interest in learning basic semiconductor physics and how this physics applies to optical sensors. Chem 102 corequisite could help, but not mandatory.
• Interest in engaging in interdisciplinary research
• Curiosity and good communication skills
• Graphic design / interest in making scientific figures could help undergraduate make a good contribution to my review paper.
• No prior wetlab experience necessary, though expect to be doing chemical synthesis

Time commitment: 4 – 6 hours a week; encouraged to also attend weekly group meetings.

About

My research focuses on security and resilience for high-performance computing (HPC) systems, network traffic analysis, and agentic AI. I lead the AICyberLake project, an initiative focused on building an operational security data lake from supercomputer logs to study and mitigate cyberthreats. I look forward to introducing early undergrads to data-driven security research and helping them build strong analytical foundations.

Description of possible projects

Students will work with real-world operational security log data collected from the DeltaAI supercomputing ecosystem. The goal of this project is to develop a deep understanding of emerging network protocols and system behaviors under failure and attack conditions. Students will learn how to parse structured and unstructured logs (e.g., Zeek network data or system authentication logs) and use Python tools to derive statistical insights.

Desired skills and time commitment

C++/Python/Networking/Statistics

About

Laura Bickart is a fourth year chemistry PhD student working in Professor Jeff Chan’s lab at UIUC. Her primary focus is on the development of on-demand fluorogenic probes to investigate the effect of nitric oxide on endosomal dynamics. Like much of the work in the Chan lab, this project lies at the interface of organic chemistry and chemical biology, and utilizes a wide variety of skillsets from synthesis and spectroscopy, to cell culturing and imaging. Prior to starting at UIUC Laura received her Bachelor’s degree in chemistry from Smith College, where she worked in Professor Dave Gorin’s lab synthesizing DNA Targeting Small-Molecule Catalysts. Laura loves teaching, and is very excited to mentor driven (and hopefully equally as excited) students!

Description of possible projects

Possible projects would be offshoots of current thesis work on endosomally targeted fluorogenic probes. For mentees seeking research experience in synthetic organic chemistry, the work would be centered around the synthesis of a panel of novel small molecule probes with improved fluorescent properties compared to currently available endosome trackers. This project would entail researching synthetic pathways, running and purifying reactions, and analyzing spectroscopic data. Synthesis is notoriously time consuming work, so after the initial 2-3 weeks of training and literature review it is likely that one would spend an entire semester making 1-3 versions of the final target molecule.

For those more interested in biology and imaging applications, the project would be focused on taking any completed molecules and assessing their performance in vitro, with an end goal of obtaining confocal images showing on demand fluorescent turn on localized to the endosomal compartment in cells. The timeline of this specific portion of the project is highly dependent on whether the synthesis of any final molecules has been completed, so it is suggested that students interested in this project also be open to doing synthesis. As with the synthesis based project, the first 2-3 weeks will be spent completing required training, as well as reviewing the literature to ensure solid foundational knowledge. Assuming at least one version of the molecule is completed, basic in vitro experiments (ex. absorbance and fluorescence assays) would likely take 1-2 months, and the rest of the time would be spent optimizing the cell and imaging experiments.

While not a requirement, it is encouraged that students have, or are currently taking organic chemistry, as a basic understanding of nomenclature and reactivity are integral to the project. Majors or minors in chemistry, biology, or other related fields would be beneficial to an applicant. However above all else, work ethic, curiosity, and commitment to safe lab practices are valued in a potential mentee.

Desired skills and time commitment

As mentioned previously it would be beneficial if the student has taken or is currently taking organic chemistry. However if this is not the case, the student should be willing to spend some time in the beginning meeting with me one on one, and studying independently to learn organic nomenclature and basic reactivity trends. This will ensure that they are both able to fully understand the project conceptually, and understand the hazards and potential reactivity of chemicals that they are working with.

It is important that the student be willing to take initiative. There will of course be extensive training in the beginning and continued guidance throughout, but the student should strive to become proficient in brainstorming experiment ideas, finding procedures in the literature, and setting up experiments independently. On this note, the student need not have prior experience with experimental planning, excitement to learn and curiosity are the most important thing, the rest will follow!

As far as time commitment is concerned, baring the occasional time conflict the student is expected to attend weekly group meetings, which are held on Thursday evenings from approximately 5-7 pm. Students should expect to spend approximately 10 hours per week in the lab. The schedule is very flexible however, and it is completely understandable if a student has to spend less time in a given week due to exams or any other extenuating circumstances.

About

Mauro Martinez is a third-year PhD researcher in Crop Sciences working at the National Soybean Research Center (NSRC) in UIUC. His research focuses on integrating crop growth modeling, environmental characterization, photosynthesis measurements, and machine learning into breeding pipelines to enhance genomic predictions and improve grain yield. He advocates for interdisciplinary collaborations, as it helps build innovative solutions to global challenges.

Description of possible projects

There will be two projects ongoing: 1) Soybean Crossing and Photosynthetic Trait Evaluation, and 2) AI Integration for APSIM Crop Modeling.

Project 1: Soybean Crossing and Photosynthetic Trait Evaluation

This project focuses on evaluating photosynthetic performance across parental soybean lines and selected progeny. Students will gain hands-on experience in plant breeding, greenhouse management, and physiological phenotyping. The project aims to investigate how genetic variation influences photosynthetic efficiency and related traits, with the long-term goal of identifying breeding strategies that can improve crop productivity. I am looking for students majoring in Crop Sciences, Plant Biology, Agricultural and Biological Engineering, Biology, Agronomy, or related fields. An interest in plant breeding, genetics, plant physiology, or greenhouse research is highly encouraged. Experience with greenhouse work, data analysis, or statistical software such as R is a plus, but not required. The first phase will involve learning the techniques required for controlled pollination and crossing in greenhouses (Weeks 1 to 3). During the second phase, students will assist with crossing and seed collection, as well as learn the fundamentals of photosynthesis measurements (Weeks 4 to 8). In the final phase, parental lines and progeny will be evaluated for photosynthetic traits, such as leaf mass per area, stomatal conductance, chlorophyll content, and carbon assimilation rate (Weeks 9 to 12). The final deliverables will include a curated dataset of photosynthetic measurements, documentation on crossing success and progeny development, statistical analyses, and a research report summarizing key findings and future breeding opportunities.

Project 2: AI Integration for APSIM Crop Modeling

This project explores the integration of Artificial Intelligence (AI) with the Agricultural Production Systems Simulator (APSIM), one of the most widely used crop simulation platforms. Depending on student interests, the project offers two complementary pathways: (1) AI-driven optimization of APSIM parameters using Machine Learning (ML) or Deep Learning (DL) techniques, and (2) development of AI-assisted workflows that connect APSIM with Large Language Models (LLMs) such as Claude through Model Context Protocol (MCP) servers and desktop AI environments. Students majoring or interested in Data Science, Computer Science, Agricultural Engineering, Crop Sciences, Statistics, or related disciplines are encouraged to apply. Familiarity with Python programming is highly recommended. Experience with machine learning frameworks, APIs, or crop modeling software is beneficial but not required. Students selecting the MCP/LLM integration pathway should have an interest in software development and AI-assisted scientific workflows. During the first month, students will learn APSIM fundamentals, review simulation workflows, and explore current applications of AI in crop modeling and decision support systems (Weeks 1 to 4). For the optimization pathway, the middle portion of the semester will focus on developing ML or DL models to improve parameter calibration, automate sensitivity analyses, and enhance predictive performance (Weeks 5 to 8). For the MCP/LLM integration pathway, students will investigate methods for connecting APSIM with AI assistants through MCP servers, enabling natural-language interaction, simulation management, and automated interpretation of model outputs (Weeks 5 to 8). In the final phase, students will validate their approaches through case studies, benchmark performance improvements, and document workflows for future research applications (Weeks 9 to 12). The final deliverables will include either (1) an AI-enhanced APSIM calibration and optimization framework with performance evaluation results, or (2) a functional prototype demonstrating APSIM integration with an AI assistant through MCP architecture. Additional deliverables will include technical documentation, reproducible code repositories, and a final presentation summarizing the project’s outcomes and future development opportunities.

Desired skills and time commitment

For project 1, no prior research experience is required, and students will be trained in greenhouse operations, soybean crossing techniques, and photosynthetic measurements. We welcome students with interests in plant biology, crop sciences, genetics, agriculture, or hands-on research involving living plants. For project 2, no prior research experience is required. Students with interests in artificial intelligence, software development, crop modeling, and scientific workflows are encouraged to apply; experience with Python, R, SQL, APIs, or Git is helpful but not required. Understanding of client-server architecture, familiarity with MCP concepts, and knowledge of JSON, APIs, or web services are a plus. Students should be prepared to commit approximately 4–8 hours per week throughout the semester for both projects.

About

Lucille Violette Dentice is a third-year Ph.D. student in MechSE at the University of Illinois. She earned her B.S. in Materials Science and Engineering from the Georgia Institute of Technology, with a concentration in structural and functional materials. Lucille studies the complex interactions between alloy processing and corrosion, with a focus on chloride-induced stress corrosion cracking of stainless steel weldments. She is deeply interested in identifying and exploring how stress corrosion mechanisms interact with bulk materials defects and elemental segregation. Her research primarily informs design and corrosion mitigation in alloys used in spent nuclear fuel canisters. Lucille is a fellow of the DOE Office of Nuclear Energy’s University Nuclear Leadership Program.

Description of possible projects

The basic premise of the project is as follows. The pieces that we would like mentee assistance with are denoted by asterisks:

• We are studying the response of 5 different heats of 304L stainless steel to stress corrosion cracking. SCC testing is carried out off-site by collaborators; however, we want to characterize our materials in their as-received state to correlate microstructural and elemental differences to SCC behavior.
• Material needs to be mechanically sectioned and polished to a smooth finish to acquire well-distributed and accurate EDS data. This process can be time-consuming and is difficult to automate.
• We need a large sample of SEM-EDS data across different samples and regions of the material to be able to generalize our observation. SEM-EDS mapping is an excellent introduction to electron microscopy.
• If we identify heats of material with strong compositional segregation, then the project focus may shift to performing electrochemical characterization (potentiodynamic polarization, Kelvin probe force microscopy, possibly other techniques). Any mentees working on the project would be briefed on the required techniques and given the option to remain on the SEM-EDS work if they desired.

Desired skills and time commitment

No skills are outright required; I would be thrilled to coach a student in any of the techniques that we employ. That said, mentees will be expected to:

1. Communicate early and often about hurdles, questions, and concerns that may arise,
2. Demonstrate a proficiency with ImageJ and Python for image analysis,
3. Learn metallographic sectioning and polishing techniques,
4. Use the MRL’s Axia SEM to collect SEM-EDS data on samples,
5. Prepare short presentations with professionally-formatted figures on their research progress.

I am happy to discuss time commitment on a mentee-by-mentee basis, depending on their courseload and perceived availability.

About

Kimberly is originally from Costa Rica. She received her B.S. in Geography with emphasis in Land Planning, from the National University of Costa Rica. After six years working in the private sector, she returned to academia to pursue graduate studies. During her M.S. at the University of Puerto Rico, she evaluated soil fertility indicators to improve phosphorus recommendations for tropical pastures. As a Ph.D. student at the U of I, Kimberly studies greenhouse gas emissions in organic grain systems, combining systematic literature reviews, field experiments, spatial analysis, and statistical modeling to better understand nitrogen cycle in these systems. She is also interested in applying artificial intelligence and machine learning to improve GHG estimates and support more sustainable agricultural systems.

Description of possible projects

Project 1. Characterizing soil carbon pools in organic grain systems Soil carbon plays a central role in nutrient cycling and greenhouse gas (GHG) production in agricultural systems. This project will investigate how different soil carbon pools vary across organically managed grain fields in Illinois. The undergraduate researcher will receive hands-on training in soil laboratory techniques, including soil sample preparation, particulate organic matter (POM) fractionation, total carbon and nitrogen analysis by dry combustion, and dissolved organic carbon (DOC) quantification. Throughout the semester, the student will learn laboratory quality control, data management, and the scientific principles underlying soil carbon cycling. The project will contribute directly to ongoing research examining how soil characteristics influence greenhouse gas emissions in organic agricultural systems while providing the student with practical experience in soil science research. Students will participate in weekly meetings with the mentor to discuss research progress, interpret results, and gain exposure to the process of conducting graduate-level research.

Recommended Background/Skills

• Majors/Minors: Natural Resources and Environmental Sciences, Crop Sciences, Environmental Engineering, Agricultural and Biological Engineering, Earth, Society, and Environmental Sustainability, Chemistry or related STEM disciplines.
• Skills: Interest in soil science, agriculture, or environmental sustainability. Careful attention to detail. Ability to follow laboratory protocols (SOP). Basic spreadsheet skills (Excel). Curiosity, patience and willingness to learn.

Timeline

Weeks 1-2

• Laboratory safety.
• Project introduction.
• Soil sample preparation.
• Soil pH and texture analyses.

Weeks 3-8

• POM fractionation.
• Total carbon and nitrogen analyses.
• Data organization and quality control.

Weeks 9-13

• DOC analysis.
• Continue data management.

Weeks 14-16

• Data verification.
• Research summary.
• End-of-semester presentation.

Deliverables

• Complete laboratory analyses for assigned samples.
• Organized quality-controlled dataset.
• Short research summary.
• Final presentation during a lab meeting or undergraduate symposium.

Project 2. Investigating soil nitrogen dynamics in organic grain systems. Nitrogen availability regulates crop productivity while influencing nitrous oxide (N₂O), one of agriculture’s most important greenhouse gases (GHG). This project focuses on measuring soil inorganic nitrogen pools in organically managed grain fields in Illinois. The undergraduate researcher will learn laboratory methods to quantify ammonium (NH₄⁺), nitrite (NO2⁻), and nitrate (NO₃⁻), assist with soil sample preparation, participate in data quality control, and explore how soil nitrogen dynamics relate to greenhouse gas emissions. The student will gain experience with laboratory techniques commonly used in soil biogeochemistry while contributing to active ongoing research examining how soil characteristics influence greenhouse gas emissions in organic agricultural systems. Students will participate in weekly meetings with the mentor to discuss research progress, interpret results, and gain exposure to the process of conducting graduate-level research.

Recommended Background/Skills

• Majors/Minors: Natural Resources and Environmental Sciences, Crop Sciences, Environmental Engineering, Agricultural and Biological Engineering, Earth, Society, and Environmental Sustainability, Chemistry or related STEM disciplines.
• Skills: Interest in soil science, biogeochemistry, agriculture, or environmental sustainability. Careful attention to detail. Ability to follow laboratory protocols (SOP). Basic spreadsheet skills (Excel). Previous laboratory experience is helpful but not required. Curiosity, patience and willingness to learn.

Timeline

Weeks 1–2

• Laboratory safety.
• Project introduction.
• Soil sample preparation.

Weeks 3–6

• Learning inorganic nitrogen extraction procedures.
• NH₄⁺, NO2⁻, and NO₃⁻ analyses.
• Quality control.

Weeks 7–12

• Continue laboratory analyses.
• Data entry and organization.
• Preliminary interpretation of results.

Weeks 13–16

• Finalize dataset.
• Prepare figures and summary.
• End-of-semester presentation.

Deliverables

• Complete inorganic nitrogen analyses for assigned samples.
• Quality-controlled laboratory dataset.
• Short research summary.
• Final presentation during a lab meeting or undergraduate symposium.

Desired skills and time commitment

No previous research or laboratory experience is required. Students should be curious, motivated to learn, and interested in environmental science, agriculture, or soil science. Attention to detail, organization, and the ability to carefully follow laboratory protocols are important. Basic experience with spreadsheets (e.g., Excel or Google Sheets) is helpful but not required. Students can expect to commit 8–10 hours per week throughout the semester. This includes laboratory work, weekly meetings with the mentor, data organization, and participation in discussions about research progress and scientific methods. By the end of the semester, students will have developed hands-on laboratory skills, experience working with research data, and a better understanding of how scientific research contributes to sustainable agriculture.

About

Gabriel is a 6th-year PhD Candidate in Electrical and Computer Engineering working under the guidance of Prof. Jennifer Bernhard. His current research focuses on the design, characterization, and practical limitations of orbital angular momentum (OAM) generating antennas. Before joining UIUC, Gabriel completed his B.Sc. degree in Electrical Engineering at the University of Puerto Rico – Mayagüez Campus. Gabriel is excited to mentor undergraduate students seeking to learn more about antenna engineering and automated microwave measurements.

Description of possible projects

Project: A Graphical User Interface for Automated Antenna Range Measurements in Anechoic Chamber Environment

Microwave antennas are essential to the development of wireless systems, allowing our devices to interact with each other over vast distances. Unfortunately, current antenna measurement setups require considerable human effort and cumbersome calibration procedures. Our group has developed a variety of scripts to automate parts of this process; however, we are looking for eager researchers interested in consolidating these efforts on a single user-friendly program.

The student will develop a MATLAB or Python-based graphical user interface (GUI) that automates the antenna radiation pattern measurement process. We have already developed rudimentary MATLAB scripts to automate parts of these measurements. However, current scripts are highly specialized to specific antennas under test (AUTs) and do not have a GUI.

All majors are encouraged, but preference will be given to ECE, CS, and Physics majors. Students should be comfortable with either MATLAB or Python.

Tentative timeline

Weeks 1-2: We will introduce the student to current antenna pattern measurement setups and train them in equipment use. Weeks 3-4: Student will review existing code and receive training on VISA communication protocols. Weeks 5-7: Student will work on consolidating existing scripts to interface with laboratory equipment and take general antenna pattern measurements. Weeks 8-11: Student will develop a GUI and refine measurement scripts. Week 12: Student will demo the GUI to the Antenna Research team.

Desired skills and time commitment

MATLAB or Python experience required (CS 101). Student will meet with their mentor once a week, or more if needed, and is expected to commit to at least 5 hours each week.

About

Hello! I am Jordan Ouellet, a senior research engineer at the Newmark Structural Engineering Laboratory. I recently graduated from UIUC with a Ph.D. in transportation and construction materials engineering. As a research engineer, I manage the lab space and help students achieve their research goals. I am originally from Montreal, Canada, and I love breaking things for science!

Description of possible projects

Our lab (that breaks materials and structures for a living) is transitioning to a digital asset management system with barcodes, to help track the equipment and allow rentals. Since the lab relies on high precision and accuracy, we want to be able to carefully monitor and know when an item needs attention. We are currently using an online platform, but we would need someone to help us configure it and help students use it. We may be open to other platforms as well! This is a chance to familiarize yourself with state-of-the-art lab equipment and learn more on civil and structural engineering research. This project may contain side quests like 3D printing and part design! Learn more on nsel.cee.illinois.edu

Desired skills and time commitment

We are looking for someone who is organized, hands-on, detail-oriented, and interested in technology and engineering research. Experience with databases, spreadsheets, or inventory management software is helpful but not required. Students interested in civil, structural, mechanical, material, or computer engineering are encouraged to apply!

About

Aagam Jain is a Master’s student in Aerospace Engineering at the UofI, an Analog Astronaut at HI-SEAS, Hawaii — home to NASA’s five successful long-duration Mars simulation missions — and a recent Aerospace Engineering graduate from VIT University, India. As a researcher at the Space Applications Centre (SAC) and the Vikram Sarabhai Space Centre (VSSC), both part of the Indian Space Research Organisation (ISRO), he gained hands-on experience developing aerospace solutions in materials and human factors. His research now focuses on aerospace innovation and human factors to enhance astronauts’ performance and resilience in extreme environments — driven by a vision for interplanetary civilization. He serves as a Research Mentor for AE 298 at UofI, and is a core member of ARKASA, where the team designs advanced lunar and Martian habitats by bridging ancient building techniques with modern space technology. His current thesis work involves building a neuroadaptive biofeedback system using AR/VR to support astronaut mental health — a field he’s excited to pioneer here. Aagam is looking for motivated undergraduates to join him on these projects, whether their interests lie in smart materials and SolidWorks or psychology and human-centered design — he’s eager to help mentees build technical skills and confidence as they work together to reinvent the future of space exploration.

Description of possible projects

Project 1: Closed-Loop Neuroadaptive VR Biofeedback — Human Subjects Data Collection

Now that the study has full approval, we’re moving into actual data collection: 60 participants running a ~50 minute VR session analog to a spaceflight scenario, wearing EEG/ECG sensors while a neuroadaptive system monitors their “Focus Index” in real time. A mentee would help run sessions — screening/scheduling participants, assisting with equipment setup (Meta Quest 3, OpenBCI array), administering pre/post surveys via Qualtrics, and keeping data organized and de-identified.

Project 2: EEG/ECG Signal Processing for Cognitive Workload Detection

This is the analysis side — taking the raw physiological data and turning it into something meaningful. Work involves MATLAB/EEGLAB, computing power spectral density (Welch’s method), and helping validate the theta/alpha-based Focus Index against participants’ self-reported stress and workload scores. Skills: MATLAB/EEGLAB and background reading.

Desired skills and time commitment

Looking for someone reliable and detail-oriented — no research background required. Expect roughly 3–5 hours/week, with some flexibility needed since lab sessions occasionally fall outside normal class hours. Interest in aerospace, human factors, CS, or neuroscience is a plus

About

Limo Schissler is a third-year Ph.D. student in the Structural Engineering area within Civil Engineering, and they also received B.S. and M.S. degrees from CEE at Illinois. Limo’s research with Prof. Larry Fahnestock and Prof. James LaFave explores innovative methods for bridge inspection and assessment, such as laser scan data analysis for characterizing steel bridge girders damaged by over-height vehicle collisions. Outside of engineering, Limo enjoys playing trumpet in musical ensembles and currently serves on the graduate teaching staff for the Marching Illini band.

Description of possible projects

Project title: Media representation of low-clearance bridges notorious for damaging strikes

Project description

Low-clearance bridges are susceptible to over-height vehicle strikes. Despite warning systems and signage, some bridges are struck frequently and become notorious for these impact events. Examples include the “11 foot 8 Can Opener Bridge” in Durham, NC, the “Big Penny” bridge in Lansing, MI, and the Storrow Drive bridges in Boston, MA. While some low-clearance bridges are simply regarded as inconvenient or hazardous, others become local icons and inspire celebrations such as community parties, local beers, and humorous videos. Whether viewed positively or negatively, a notorious, frequently-struck low-clearance bridge is a local landmark that contributes to a unique sense of place.

This project will investigate how notorious, frequently-struck bridges contribute to place identity. We will examine a variety of case studies and compare quantitative and qualitative factors to understand the relationship between the bridge and its surroundings (both physical and social). Data sources for this project may include both popular and professional / technical media, such as news broadcasts, newspaper articles, bridge “fan” websites, social media, the National Bridge Inventory database, and other online sources. A corpus linguistic analysis of words and phrases used in these regards will be conducted to help compare and contrast media discourse across different cases.

This research is interdisciplinary and includes concepts from civil engineering and urban studies. Suggestions and insights from undergraduate researchers will significantly contribute to the project’s success. This should be an enjoyable and engaging opportunity to explore bridges and infrastructure from a different perspective compared to traditional engineering coursework.

Recommended majors: This project will most likely appeal to civil engineering and/or urban planning students, but anyone interested in bridges and human factors is welcome.

Required qualifications: familiarity with major news outlets in the U.S., familiarity with social media, interest in an interdisciplinary project

Optional qualifications: hobby or interest in the arts/humanities (e.g., literature, linguistics, cultural studies, visual art, music, film/media, history, etc.)

Timeline

• Weeks 1-2: Introduce research topic and framework
• Weeks 3-4: Identify suitable bridges for the study
• Weeks 5-10: Collect and analyze data
• Weeks 11-12: Synthesize overall trends in data; finalize deliverables

Deliverables: Annotated list of data sources, written summary describing insights learned, URSA final presentation

Desired skills and time commitment

Desired skills: reading, organization, enthusiasm, and critical thinking Time commitment: approximately 5 hours per week

About

Saumitra Rai is a Ph.D. student in Environmental Engineering at the University of Illinois Urbana–Champaign. He received his B.E. (Hons.) in Civil Engineering from BITS Pilani and his M.S. in Environmental Engineering from the University of Illinois Urbana–Champaign. His research focuses on developing process-based and geospatial modeling frameworks to support wastewater infrastructure planning under data scarcity. He is particularly interested in integrating wastewater process modeling, spatial analysis, and uncertainty analysis to evaluate the economic, environmental, and public-health implications of infrastructure decisions. His doctoral research examines wastewater systems across multiple spatial scales, including the effects of industrial decarbonization strategies on water resource recovery facilities, the planning of centralized and decentralized sanitation systems in rapidly growing cities, and the estimation of GHG emissions from wastewater systems across the United States. More broadly, his work seeks to translate incomplete infrastructure, population, and operational data into scalable and context-sensitive decision-support tools that identify where wastewater interventions are most likely to be cost-effective, environmentally beneficial, and practically implementable.

Description of possible projects

Project Title: Evaluating Solar-Plus-Storage for Resilient Wastewater Treatment

Project Description: Wastewater treatment plants (WWTPs) require continuous electricity for pumping, aeration, disinfection, and other essential operations. This project will investigate whether solar photovoltaic systems and battery storage could help maintain critical treatment functions during grid outages. The student will develop a computational screening model for a selected group of facilities in the United States, South Africa, or both, depending on data availability.

The project will combine wastewater-facility characteristics, estimated electricity demand, historical outage or load-shedding patterns, and location-specific solar-resource data. The student will simulate how different solar and battery sizes perform during outages and identify facilities where these systems could provide the greatest resilience benefit. The semester project will focus on a manageable pilot region rather than a complete national assessment. South Africa provides a particularly relevant application because load-shedding has produced frequent, geographically variable power interruptions that can be reconstructed at relatively fine temporal and spatial scales.

Preferred Background and Skills: Students in Environmental, Civil, Electrical, or Computer Engineering; Data Science; or related majors are encouraged to apply. Basic Python programming and comfort working with tabular data are strongly preferred. Use of AI for coding is highly encouraged, so students with no programming experience are welcome. More experience with pandas, data visualization, GIS or geopandas, energy-system modeling, or wastewater treatment is helpful but not required. The student should be comfortable learning new datasets and working independently with guidance.

Approximate Timeline

Weeks 1–3: Literature review, selection of study area, and identification of facility, outage, and solar datasets. Weeks 4–7: Data cleaning, spatial matching, and estimation of facility electricity demand. Weeks 8–11: Development of a simplified hourly solar-and-battery simulation. Weeks 12–14: Scenario analysis, visualization, and identification of priority facilities. Weeks 15–16: Final report, documented code, and presentation.

Expected Deliverables: The student will produce a cleaned pilot dataset, reproducible Python code (Use of AI for coding is highly encouraged), maps or figures summarizing the results, and a short technical report and presentation describing the methodology, findings, limitations, and opportunities for future work.

Time Commitment: Approximately 4–6 hours per week over one semester. Working hours are very flexible since the work is computational. From September through November, I will be a visiting researcher at Eawag in Switzerland, so regular mentoring meetings during this period will be held via Zoom. I will be available for weekly meetings and additional virtual meetings as needed. My PI, Prof. Jeremy Guest, and other members of our research group will also be available to provide in-person guidance as needed. I will be available for in-person meetings beginning November 29.

Desired skills and time commitment

No specific coursework or prior research experience is required. Applicants should be curious, dependable, and willing to learn basic Python-based data analysis and computational modeling. Familiarity with spreadsheets, programming, GIS, or environmental and energy systems is helpful but not expected.

About

Tony Hong is an M.S. student in Computer Science at the University of Illinois Urbana-Champaign, where he is advised by Professor Fan Lai. He received his B.S. in Electrical Engineering and Computer Sciences from UC Berkeley, where he worked with Professor Ion Stoica. His research spans machine learning systems, distributed systems, operating systems, and storage systems, with a particular focus on efficient infrastructure for AI agents and large language models. His current research centers on AI agent serving. He is developing an attention-aware serving system that jointly optimizes distributed prefill, KV-cache orchestration, and sparse decoding for long-running agents. More broadly, he is interested in designing systems abstractions and resource-management techniques for the dynamic, stateful, and data-dependent nature of agentic workloads. His research has appeared at leading systems conferences, including SIGCOMM, OSDI, SOSP, and ASPLOS.

Description of possible projects

Project 1: Just-in-Time Execution for Multi-Agent LLM Applications

Modern AI applications increasingly consist of multiple LLM agents connected by data dependencies. For example, one agent may generate a proposed solution, another may verify it, and a third may combine their outputs. Existing LLM serving systems generally execute each model invocation independently: a downstream agent waits for its entire input to arrive before processing it. This serialization can substantially increase end-to-end latency.

Magnet treats the complete agent workflow as a dataflow graph rather than a collection of isolated LLM calls. It begins downstream context processing while upstream agents are still generating, allowing computation from dependent stages to overlap. Its main mechanisms include just-in-time prefilling, which selects when downstream processing should begin; parallel prefilling, which processes inputs from independent branches concurrently; causal reconciliation, which repairs missing dependencies introduced by parallel execution; and dataflow-aware resource allocation across heterogeneous agents.

A semester project would focus on one well-defined component of Magnet rather than implementing the entire system. Possible directions include building a trace-driven simulator for just-in-time prefill scheduling, collecting execution traces from representative multi-agent applications, prototyping a lightweight dataflow scheduler on top of an existing inference engine, or evaluating policies for determining when downstream prefilling should begin.

Suggested background

Required

• Major or substantial coursework in Computer Science, Electrical Engineering, or a related field
• Proficiency in Python
• Familiarity with basic systems concepts, such as scheduling, concurrency, and latency/throughput tradeoffs

Preferred but not required

• Experience with PyTorch or transformer inference
• Familiarity with LLM serving systems such as vLLM or SGLang
• Experience with distributed systems, CUDA, or performance profiling
• C++ experience for students interested in runtime implementation

Tentative timeline

• Weeks 1-3: Learn the LLM prefill and decoding execution model; run representative agent workflows and reproduce a sequential serving baseline.
• Weeks 4-6: Instrument the workflows and collect traces describing token production rates, prefill times, memory consumption, and dependency structure.
• Weeks 7-10: Implement a simulator or prototype of one scheduling mechanism, such as just-in-time prefill initiation or dataflow-aware resource allocation.
• Weeks 11-13: Evaluate the mechanism across different graph structures, model sizes, request loads, and output lengths.
• Weeks 14-15: Analyze results, document the implementation, and prepare a final report and presentation.

Expected deliverables

• A reproducible workload and measurement suite for multi-agent LLM applications
• A trace-driven simulator or working prototype integrated with an existing serving framework
• Comparisons against sequential execution and simple streaming baselines
• An evaluation of end-to-end latency, throughput, GPU utilization, and KV-cache memory consumption
• A written report and final presentation describing the design and findings

Project 2: Attention-Aware Context Orchestration for Long-Running AI Agents

Long-running AI agents accumulate large contexts containing conversation history, retrieved documents, intermediate reasoning, and tool outputs. Their KV caches can occupy substantial GPU memory. When an agent pauses for a tool call or another agent, serving systems may move its KV cache to CPU memory. When execution resumes, they typically restore most or all of that cache – even though the next query may attend strongly to only a small portion of the history. Existing systems also make prefill partitioning, KV-cache placement, restoration, sparse attention, and request batching largely independent decisions.

Conflux explores managing context according to its attention structure throughout its complete lifetime. During prefill, it partitions long contexts into semantically or attention-related regions and repairs important dependencies across partition boundaries. These regions then become persistent KV-cache management units. When an agent resumes, Conflux uses the newly available query to identify and restore only the regions that may be important. It can also batch agents that require overlapping KV regions so that data movement and computation are shared. The broader objective is to jointly reduce restoration latency, memory usage, and data-transfer overhead while improving serving throughput.

A semester project would investigate one portion of this lifecycle. A student could characterize attention locality in agent traces, implement an attention-aware KV-region simulator, prototype staged KV restoration between CPU and GPU, or design a scheduler that batches requests based on overlap in their selected KV pages.

Suggested background

Required

• Major or substantial coursework in Computer Science, Electrical Engineering, Data Science, or a related field
• Proficiency in Python
• Understanding of basic data structures and computer systems concepts

Preferred but not required

• Familiarity with transformer attention and KV caching
• Experience with PyTorch, vLLM, FlashAttention, or FlashInfer
• Coursework in operating systems, computer architecture, distributed systems, or machine learning systems
• Experience with CUDA, C++, memory profiling, or GPU performance analysis

Tentative timeline

• Weeks 1-3: Learn transformer attention, KV caching, paged memory management, and the behavior of agents that pause and resume.
• Weeks 4-6: Collect or analyze attention and KV-access traces from long-context agent workloads; quantify locality, working-set size, and overlap across requests.
• Weeks 7-10: Implement one mechanism, such as attention-aware region construction, staged sparse restoration, or overlap-aware request batching.
• Weeks 11-13: Integrate the mechanism into a simulator or small serving prototype and compare it with full-cache restoration.
• Weeks 14-15: Evaluate tradeoffs among restoration latency, transfer volume, GPU memory usage, throughput, and model-output quality; prepare the final report and presentation.

Expected deliverables

• A dataset or profiler capturing attention and KV-cache access patterns in agent workloads
• A simulator or prototype of an attention-aware context-management mechanism
• Comparisons with full-cache restoration and request-independent sparse restoration
• An evaluation of CPU-GPU transfer volume, resume latency, memory consumption, throughput, and accuracy
• A documented codebase, written report, and final presentation

Desired skills and time commitment

Applicants should be comfortable programming in Python and have completed, or be currently taking, an introductory programming or data structures course. No prior research experience, machine learning coursework, or knowledge of LLM systems is required. We are primarily looking for students who are curious about computer systems and AI, willing to learn unfamiliar tools, and comfortable debugging code and discussing their progress.

Helpful but optional experience includes

• Basic familiarity with PyTorch or machine learning
• Experience using Linux, Git, or command-line tools
• Interest in systems topics such as scheduling, concurrency, memory management, or distributed computing
• Familiarity with C++ for students interested in lower-level implementation

Students will receive guidance on the relevant concepts, including transformer inference, attention, KV caching, and LLM serving. Initial tasks will be scoped around accessible activities such as running existing code, collecting and analyzing performance measurements, implementing scheduling policies in Python, and building small prototypes.

We expect a commitment of approximately 6-8 hours per week throughout the semester, including a weekly research meeting. Students should be able to work consistently, communicate when they encounter difficulties, document their experiments, and present their progress periodically. No previous publications or independent research projects are expected.

About

Hi, I’m Prachit Gupta, a Master’s student in Mechanical Science and Engineering at UIUC and a graduate researcher at the N ACXIS Lab in Aerospace Engineering.

My research focuses on making autonomous systems safe, trustworthy, and capable of behaving in ways that mimic human-like decision making. As autonomous systems become increasingly common around us, I believe it is paramount that they remain provably safe and trustworthy for the people who interact with them, even if they have no understanding of how the underlying technology works. More recently, I have been exploring how foundation models such as Large Language Models (LLMs) can help realise that broader goal.

I primarily work with aerial drones and am expanding these ideas to a variety of robotic platforms, including flapping-wing drones, spider-inspired robots, hexarotors, and ground vehicles. My research is inherently interdisciplinary, bringing together ideas from mechanical engineering, aerospace engineering, controls, robotics, machine learning, and computer science.

Before joining UIUC, I was part of a student technical team developing UAVs for search-and-rescue applications, where we presented our work at international forums across Europe and the United States and received several international recognitions and awards. Looking back, much of my own journey has been shaped by working within mentor–mentee teams and receiving early exposure to research. Through URSA, I hope to provide students with that same first exposure to hands-on robotics research and help them discover just how exciting it is to build and experiment with real autonomous systems. You can check out my website for more details of these past projects : https://prachitgupta.github.io/personal_website/

Description of possible projects

Safe and Trustworthy Language-Guided Autonomous Robotics

This project explores how recent advances in foundation models, particularly Large Language Models (LLMs), can be used safely on real robotic systems.

Our current research investigates how LLMs can interpret high-level natural language instructions and generate mission-level waypoints for autonomous quadcopters. These waypoints are converted into dynamically feasible trajectories that robots can execute safely. Since modern AI models are inherently uncertain, an important part of this project is estimating how reliable those predictions are and ensuring that robotic systems remain within mathematically guaranteed safety bounds even when the model makes imperfect decisions.

Although the current implementation focuses on quadcopters, the long-term goal is to extend these ideas to a broader class of robotic platforms, including flapping-wing drones, spider-inspired robots, hexarotors, and ground vehicles. The exact research direction may evolve throughout the semester depending on project progress and ongoing research developments.

Tentative Timeline

Week 1 — Lab Onboarding & Hardware Familiarization

Introduction to the research lab and available robotic platforms. Overview of drones, flapping-wing robots, hexarotors, spider-inspired robots, and supporting hardware. Setting up the development environment and becoming familiar with the overall research workflow.

Week 2 — Software Setup & Reproducing Existing Results

Install required software packages and dependencies. Learn the existing software pipeline. Reproduce baseline simulation results to gain familiarity with the current implementation.

Week 3 — Theory & Background

Guided reading of selected papers and documentation. Introduction to the mathematical foundations required for the project, including basic control theory, uncertainty estimation, and the intuition behind conformal prediction and contraction-based safety methods. Discussion sessions to connect the theoretical concepts with the software implementation.

Week 4 Onwards — Research Contribution & Hardware Experiments

Students will begin contributing to the project based on their interests and progress. Possible activities include:

Conducting hardware experiments on autonomous robotic platforms. Collecting and analyzing experimental data. Testing and validating the existing pipeline. Implementing improvements to the software stack. Exploring extensions of the framework to additional robotic platforms where appropriate.

Since this is an active research project, the exact direction during the latter half of the semester may evolve based on experimental results and ongoing research discussions.

Expected Deliverables

Successfully reproduce baseline simulation results. Contribute to hardware experiments. Document implementation and experimental findings. Present results through a final report, poster, demonstration, or research presentation.

Desired skills and time commitment

The most important qualification is genuine curiosity about robotics and a willingness to build, debug, and learn through hands-on experimentation. Prior research experience is not required.

Helpful (but not required)

1. Basic Python programming.
2. Comfort with mathematical equations and basic linear algebra.
3. Interest in robotics, autonomous systems, or control. Any previous hands-on hardware experience (robotics, electronics, drones, Arduino projects, etc.).
4. Experience working with drones is a plus but not expected.
5. Interest in Large Language Models, prompt engineering, AI-assisted programming, or “vibe coding.”

Students from any engineering or related STEM discipline are encouraged to apply if they enjoy learning by building real systems.

Time Commitment

Because this project involves real hardware experiments, the workload can vary from week to week depending on experimental progress and debugging requirements. Students should expect to dedicate approximately 8–10 hours per week on average, including meetings, independent work, software development, and hardware experimentation.

Some weeks may require longer experimental sessions, while others will focus more on software development, documentation, or literature review. Consistent communication, enthusiasm, and a willingness to troubleshoot real-world systems are valued far more than prior expertise.

About

Rahat is a Graduate Research Assistant in Mechanical Engineering at the University of Illinois Urbana‑Champaign, specializing in experimental thermal engineering, frost formation and removal, sensor‑driven systems, and advanced heat‑exchanger characterization. His work integrates hands‑on experimentation with electronics, instrumentation, and data‑driven analysis to develop next‑generation frost sensing and pulse electrothermal defrosting technologies. He has led the development of capacitive frost‑sensing platforms, pulse‑defrosting control hardware, and environmental chamber testbeds, combining thermal science with practical circuit design and robust data‑acquisition pipelines. His research experience spans micro/nano fabrication, uncertainty quantification, and high‑fidelity thermal measurements.

Description of possible projects

Project 1 — Multi‑Channel Pulse Electrothermal Defrosting Switch

Project Summary

Current pulse‑electrothermal defrosting experiments rely on a single‑channel electronic switch that can deliver controlled power pulses to one sample at a time. As the experimental platform transitions toward parallel‑plate defrosting, the system needs to support two or more independently addressable heating channels. This project focuses on designing, prototyping, and validating a multi‑output electronic switching module capable of distributing pulsed power to multiple samples with precise timing, isolation, and safety.

Student Fit: Electrical Engineering, Mechanical Engineering with electronics interest

Technical Objectives

1. Extend the existing single‑channel switch to multi‑channel architecture
2. Ensure independent control of each output (timing, duty cycle, pulse width)
3. Integrate current/voltage sensing for each channel
4. Maintain electrical isolation and thermal safety
5. Validate performance using resistive dummy loads, then transition to frost samples

Proposed timeline

The mentee will begin by reviewing the existing single‑channel switch design, power requirements, and pulse characteristics (Weeks 1–2), then move into designing a multi‑channel circuit schematic, selecting switching components, and running simulations in LTspice or Multisim (Weeks 3–5). They will proceed to PCB layout, fabrication, and prototype assembly (Weeks 6–8), followed by developing microcontroller‑based control firmware and performing bench‑testing with dummy loads (Weeks 6–8). After initial validation, the student will integrate the module with the parallel‑plate frost‑defrosting setup and evaluate simultaneous and sequential pulsing performance (Weeks 9–10), concluding with documentation, PCB files, firmware, and a final performance summary (Weeks 11–12).

Project 2 — Multiplexed Frost Sensing System for Multi‑Sensor Data Acquisition

Project Summary

The frost sensing platform currently reads data from individual capacitive/impedance sensors one at a time. As the system expands to sensor arrays, the data acquisition pipeline must support multiplexing — either simultaneous or sequential — while maintaining signal integrity and timing accuracy. This project aims to design a multiplexed sensing architecture using analog/digital multiplexers, microcontroller firmware, and signal‑conditioning circuits to acquire frost‑growth data from multiple sensors.

Student Fit

Primary: Electrical Engineering, Mechanical Engineering

Technical Objectives

1. Implement analog or digital multiplexing for 4–8 sensors
2. Maintain stable capacitance/impedance measurements under frost conditions
3. Develop firmware for synchronized sampling
4. Integrate data logging and timestamping
5. Validate system using controlled frost‑growth experiments.

Proposed Timeline

The mentee will start by studying the current single‑sensor frost‑sensing setup and defining sampling rate, resolution, and multiplexing requirements (Weeks 1–2), then prototype an analog/digital multiplexing circuit, implement signal conditioning, and begin microcontroller firmware development for channel switching and data acquisition (Weeks 3–5). They will integrate the multiplexing module with the existing sensing hardware and characterize noise, drift, and cross‑talk while refining sampling timing (Weeks 6–8). Once stable operation is achieved, the student will run frost‑growth experiments using a multi‑sensor array to compare simultaneous versus sequential acquisition (Weeks 9–10), finishing with a complete set of deliverables including firmware, wiring diagrams, datasets, and documentation for future expansion (Weeks 11–12).

Desired skills and time commitment

Required Skills and time commitment for Mentees

Project 1: Basic circuit analysis (Ohm’s law, power calculations, transient behavior), understanding of PWM, pulse timing, and digital control signals, experience with data acquisition or sensor integration; Weekly 4-5 hours

Project 2: Basic understanding of analog/digital signals, ability to implement signal conditioning (filters, grounding, shielding), comfort with data acquisition and logging; Weekly 4-6 hours