The 14 currently listed mentors are for the Fall 2024 semester.
If you have any questions about the information below, please reach out to the URSA Committee at URSA-group@office365.illinois.edu.
Why become an URSA Mentor?
| Mentor | Description |
|---|---|
| Austin Lu (Electrical and Computer Engineering)  | Austin is a second-year MS student in Electrical Engineering working with Professor Andrew Singer on spatial audio and acoustic signal processing. His recent work involves developing novel tools for microphone array research and dataset collection, often with robots or 3D printed mechanisms. Description of Possible Projects: 3D printed microphone array rig/holder: To study a soundfield, we need to sample it in space and time. This basically just means placing microphones in precise locations and taking multi-channel audio recordings, which we can then use for acoustic signal processing, beamforming, and so-on. People in literature use massive aluminum robots, laser-calibration, and so-on. We are interested in more convenient methods, for instance by 3D printing structures that hold many microphones in a desired configuration. In this project, students (preferably in ME or SED) will design such structures. We have a 3D printer in-house that’ll make design iteration as fast as possible. If time permits, the students will take recordings from their structures with the lab’s massive multichannel audio interface and microphones. Desired Skills and Time Commitment: Required skills: CAD (preferably Fusion), Prior experience with 3D printing (FDM/FFF) Optional: Experience with audio equipment, some knowledge of acoustics Time commitment: 3-5 hrs weekly |
| Ayush Raman (Computer Science)  | Ayush is a first year Master of Computer Science student here at UIUC. He is from Saratoga, California and is primarily interested in web development and machine learning -- and beginning to get into systems design. He is always looking to learn more through projects (you can see some at github.com/ayuram). Other than that, he plays tennis from time-to-time and is beginning to get into Valorant and Overwatch. Description of Possible Projects: Efficient Code LLM Context Caching for Debugger Agents: When working with LLMs like GitHub Copilot, ChatGPT, and Google Gemini, a concept that comes up frequently is caching larger context windows that are repeatedly used for generation. This is especially relevant in the event of a debug loop, where an agentic system is zeroing in on a particular buggy snippet of code. Dependencies (and maybe even dependents) do not need to be repeatedly computed by the LLM and can have their contexts cached. The question is: what system design would ensure that such caching remains efficient and effective? When can we afford to evict from cache? Can such a system scale effectively, and if so, what system design techniques would be needed to ensure this? Major: Computer Science General Timeline: Tentative but the general outline would go as follows: - Design the initial system on a drawing board (2 weeks) - Code the system prototype (2 week) - Debug and test the system prototype till we reach an optimal state (2 months) - Evaluate results (3 weeks) Desired Skills and Time Commitment: Experience in Python is required. Systems and ML experience is a plus. Time commitment is tentatively 10 hrs per week but I prioritize a willingness to try and learn over all else. |
| Brandon Kamiyama (Nuclear, Plasma, & Radiological Engineering)  | Brandon is a 4th year PhD student in NPRE under Prof. R. Mohan Sankaran. His current research utilizes non-equilibrium plasmas in contact with liquids for chemical conversion and biosynthesis applications. Prior to joining Mohan's lab, he got his BS in nuclear engineering at Oregon State University, specializing in experimental thermal hydraulics. Description of Possible Projects: I have one project now that needs immediate assistance that involves scaling up plasma-liquid reactors for nitrogen fixation. We are investigating multiple scale-up approaches, geometries, and techniques. We have a big milestone deadline in early January, so we need all the help we can get. We currently have 5 people working on this (postdoc/PhDs/undergrads). We need assistance in running and designing experiments primarily. NPRE or ChBE majors are preferred, but all majors are welcome to apply. I am looking for 1-2 dedicated students who have interest in staying in my group beyond this semester. In terms of a timeline, we will be designing and carrying out experiments in somewhat of an iterative manner throughout the entire semester. Desired Skills and Time Commitment: I don't require any classes or skills other than being willing to learn from day one. We will teach you everything you need to know. In terms of a time commitment, we need as much time as possible. I would prefer longer blocks of time vs. 1-2 hours per day. |
| Chandan Mahata (Agricultural and Biological Engineering)  | Chandan Mahata is a second-year doctoral student in the Department of Agricultural and Biological Engineering at the University of Illinois Urbana-Champaign (UIUC). As a research student, Mr. Mahata published 16 peer-reviewed articles in the field of bioprocess engineering and presented research outcomes at several international conferences. He has experience in supervising industrial R&D work at Dhampur Sugar Mills, India (currently Dhampur Bio-Organics Ltd) for biohydrogen production from industrial wastewater under a technology license transfer agreement. His current research focuses on precision fermentation using genetically engineered microorganisms for the production of high-value platform chemicals and the recovery of bioproducts. He is currently a peer-reviewer of several international journals, including Scientific Reports (Nature), Microbial Cell Factories (BMC), Process Safety and Environmental Protection (Elsevier), Fish Physiology and Biochemistry (Springer), 3 Biotech (Springer), and Environmental Chemistry and Ecotoxicology (Elsevier). Please visit https://www.chandanw2e.com/ for more detail Description of Possible Projects: Project 1: Production of high-value platform chemicals from renewable sources via precision fermentation using genetically engineered microorganisms. In this project, we will produce bioproducts in small scale experiments and in bioreactors. Mentees will learn fermentation process, sterile handling of microorganisms, and operation of bench-scale bioreactors. Project 2: Recovery of bioproducts from fermentation broth. Here, we will explore different downstream processes, including ion exchange chromatography, nano-filtration, and crystallization. Mentees will be trained to set-up experiments in bench-scale study. Having hands-on experience in the above projects would be advantageous to secure industrial internships. Desired Skills and Time Commitment: Students interested in wet lab experiments are welcome. Knowledge in basic statistics (design of experiments, ANOVA test), data analysis, data visualization is advantageous. For project 1, knowledge in industrial microbiology is beneficial. For project 2, chemistry and unit operations will be helpful. Meeting/working time is flexible. 5 – 10 hours per week is expected. |
| Isha Chaudhary (Computer Science)  | Isha (she/her), is a third year CS Ph.D. candidate at the Siebel school of Computing and Data Science. She studies the safety and trustworthiness of AI-driven systems such as LLMs using mathematical methods. In her recent works, she has proposed a couple of frameworks to mathematically assess the properties of fairness and reasoning in LLMs. She is looking for collaborators who can help take her aforementioned ongoing project on fairness in LLMs forward. Description of Possible Projects: This project is an extension to our framework to quantitatively certify bias in Large Language Models (LLMs) - https://quacer-b.focallab.org/. The framework is currently developed for very specialized notions of bias for only some demographic groups corresponding to the gender and racial sensitive attributes. We hope to extend it to general notions of fairness in machine learning for more sensitive attributes such as religion, disability status, age, etc, and combinations of attributes. The primary activities of the undergraduate collaborator would initially be to get involved in ideation on how the general fairness framework could be designed, and would subsequently transition to conducting experiments. Expected outcomes: The resultant framework would be an important contribution for both research and practical purposes, thus enabling the student to kickstart their career (in either research or elsewhere) at a high note. We will endeavor to publish our findings and open-source the related artifacts, so the student’s contributions can be highlighted on their profile. Moreover, the student will learn about LLMs - their algorithms, workings, and open problems in the space. They will also learn about mathematical methods for proving properties about AI models, especially LLMs. The project is expected to be experiment-heavy, hence the students can get some hands-on experience of working with GPUs and experiment with LLMs on them. Timeline: This semester-long project will begin with 2-3 weeks of onboarding, wherein the student will be explained the project and learn the important skills to start. Next 2 weeks will be a literature review phase, in which the student will study the relevant papers and understand their codebases. Following 2 weeks will consist of detailed ideation, leading to 3-4 weeks of experimentation. The remaining time will consist of documentation and open-sourcing of the project. Preferred Majors/Minors: Computer Science, Mathematics, Statistics Desired Skills and Time Commitment: Required skills: (No problem if the mentee doesn't possess these. They should just come with an intent to learn on the way!) - Knowledge of Python programming language and related libraries like Anaconda, Jupyter - Basic software development tools such as Git - Basic understanding of AI/ML including terminology, training/inferring from models, some familiarity with LLMs (relevant classes: CS 440/441/446) Desriable skills: (These are optional but good to have. Again, an intent to learn on the way would be sufficient.) - Natural Language Processing (relevant classes: CS447) - Basic probability and statistics (relevant classes: CS361 / STAT400/ STAT410) Time commitment: 8-10 hours / week |
| Jorge Maldonado (Bioengineering)  | Jorge Maldonado is a 6th-year Neuroscience Ph.D. candidate at the University of Illinois Urbana-Champaign. His research is focused on neurodegenerative diseases, particularly Alzheimer’s disease and Traumatic Brain Injury, utilizing Quantitative Phase Imaging techniques such as Spatial Light Interference Microscopy (SLIM) using a multimodal approach with Fluorescence Microscopy. Jorge main work focus on two different projects and is seeking dedicated mentees who are willing to put in the work to achieve meaningful results. Those who succeed in his projects are those who take initiative, stay committed, and strive to make significant contributions. Description of Possible Projects: Project 1: Imaging and Quantification of Traumatic Brain Injury Pathology Using Spatial Light Interference Microscopy (SLIM) In this project, we employ SLIM, a label-free optical imaging technique, to quantify morphological and optical parameters in models of Traumatic Brain Injury (TBI). The primary goals include demonstrating SLIM's compatibility with tissue clearing to enable high-resolution imaging and detailed biophysical analysis of brain sections affected by TBI. Participants will engage in tasks such as sample preparation, imaging, and data analysis, with a focus on key pathological markers like axonal injury, blood-brain barrier disruption, and glial responses. This project provides valuable experience in advanced optical imaging techniques, data interpretation, and biophysical analysis, contributing to a deeper understanding of the mechanisms underlying TBI and its effects on brain function. Project 2: Multiscale, Label-Free Imaging of Aβ Deposits in FAD5x Mouse Models Using SLIM and Tissue Clearing In this study, we aim to harness the combined power of tissue clearing and Spatial Light Interference Microscopy (SLIM) to achieve high-resolution, multiscale, and label-free imaging of amyloid-beta (Aβ) deposits in FAD5x mouse models of Alzheimer's disease (AD). This project introduces several key innovations that significantly enhance our understanding of AD pathology Desired Skills and Time Commitment: For both projects, a strong interest and basic knowledge or background in microscopy, neurodegeneration, and data analysis is essential. Familiarity with image processing, and other image analysis software will be beneficial but not required. Additionally, having a proactive attitude, attention to detail, and the ability to work independently are critical for success. Time Commitment: A minimum of 8 hours per week is required for these projects. Participants must be fully engaged, with a clear understanding that consistent effort is necessary to achieve meaningful results. This commitment will ensure that mentees can contribute effectively to the research and gain substantial experience in the field. |
| Mahdi Azizi (Agricultural and Biological Engineering)  | Mahdi is a second-year PhD student in Agricultural and Biological Engineering, working on the particle tracking velocimetry (PTV) system to study fluid flows. Description of Possible Projects: The projects are related to various steps of doing experiments with the PTV system, including but not limited to preliminary setup, capturing and processing images from cameras, calibration, testing, and running an experiment with tracer bubbles. It’s a good opportunity to familiarize yourself with PTV systems. Desired Skills and Time Commitment: Familiarity with any of the followings is an advantage: computer vision, camera calibration, programming (cpp/Matlab/python), fluid dynamics, convolutional neural networks (CNN), and collecting and analyzing data. At this stage, we will work on labeling images for neural network training. Any prior experience in labeling and CNN training is encouraged. The timing is flexible, but once we agree on your specific duties, you should be responsible to devote the required time to do it. |
| Ozgur Kara (Computer Science)  | Ozgur is a PhD student at UIUC, CS program under the supervision of Founder Professor James Rehg. His ultimate research objective is to develop controllable and explainable generative models for video applications including but not limited to text-to-video generation, video editing, long-term video generation. Beyond these, he also worked on continual learning, and inverse image problems during his previous internships. Additionally, he has had a keen interest in undertaking projects and implementing various ideas since his high school years. Now, he wants to reach out to more people at our university to engage in discussions and collaboration. Description of Possible Projects: Video Editing Mobile/Desktop Application: Text-based video editing involves providing a text prompt and a video as input, asking the model to edit the video to reflect the prompt. While several methods exist to perform this task, speed remains a significant issue. Recent work has improved the speed, making near real-time editing possible. In this project, we aim to develop a mobile/desktop application utilizing a selected video editing approach. Users will be able to record a 30-40 frame video using their phone and provide a text prompt to the application. The recorded video will be processed on a server (cloud), and the resulting video will be displayed in the application. Our goal is to release a simple yet effective application/product for this task on mobile devices/desktop. On the research side, we will explore how to achieve near real-time speed using recent advancements in diffusion models and generative models. Along the way, we will learn how these models work, their advantages and disadvantages, and implement these models in a mobile application. - For backend development, experience with cloud platforms would be highly beneficial. - For understanding how generative models work, introductory level of understanding in machine learning and Python knowledge would be helpful. - For the desktop or mobile application component, experience in application development would be helpful. - In general, experience in Python would be valuable. Candidates who are experienced in mobile development are highly encouraged to apply. Desired Skills and Time Commitment: Required: - Motivation, motivation, motivation - Willingness to learn and put in the effort Preferred Skills: - Mobile Development (Android or iOS, preferably iOS) - Experience in Python - Experience in Cloud (Backend Development connected to mobile applications) Time commitment: - Preferably 10 hours per week, but if you are motivated and enjoy the work, precise time tracking is less important. - Regular meetings at least once a week. |
| Phuong Cao (National Center for Supercomputing Applications)  | Phuong Cao is a research scientist at the National Center for Supercomputing Applications (NCSA) at the University of Illinois at Urbana-Champaign. His research interests include securing high-performance scientific computing through large-scale measurements of network protocols, designing probabilistic models to detect cyber-attacks, and synthesizing machine-assisted proofs of federated authentication protocols. His research has won a Best Paper Award, a Best Hackathon Award, and an Outstanding Mentor recognition for advising students in cybersecurity research. Phuong is a Trusted CI Fellow of the NSF Cybersecurity Center of Excellence. Description of Possible Projects: I have four projects in mind and can take at most four students. 1. Post Quantum Cryptography observatory. The problem of adopting quantum-resistant cryptographic network protocols or post-quantum cryptography (PQC) is critically important to democratizing quantum computing. The problem is urgent because practical quantum computers will break classical encryption in the next few decades. Past encrypted data has already been collected and can be decrypted in the near future. The main challenges of adopting post-quantum cryptography lie in algorithmic complexity and hardware/software/network implementation. The grand question of how existing cyberinfrastructure will support post-quantum cryptography remains unanswered. Skills needed: C/C++ or Rust programming; Network protocol; Cryptography. Deliverable: A tool to inform user of their quantum-resistant cryptography adoption rate (A website or a custom SSH server). 2. Post Quantum Data repository. Disseminate data collected by PQC observatory to NCSA's partner's such as FABRIC testbed. Skills: Jupyter notebooks, Python. Deliverable: A public Jupyter notebook contain sample data of PQC network traces. 3. Security Analytics. Testing different detection models against live network traffic collected by Zeek IDS. Skills: Statistics. Data analysis. Deliverable: A proof of concept model to perform real time detection of attacks on Zeek notice logs. 4. Synthesizing formal proof for authentication logics of federated authentication systems using Lean. Skills: compiler, z3, lean, program verification. Deliverable: A proof of concept Lean proof and corresponding implementation of one or two critical functions used in the SciTokens federated authentication library. Desired Skills and Time Commitment: Exploratory projects and low time committment. |
| Poornadithya Chandramukhi (Aerospace Engineering)  | Poornadithya Chandramukhi, is a first-year graduate student in Aerospace Engineering at UIUC. His research interests include Spaceflight Mechanics, Space Debris Management, and Satellite Attitude Determination and Control. He also has a secondary interest in solid propulsion systems and their relation to grain geometries. He has previously worked with DRDO on satellite attitude determination and DRDL on performance analysis of grain geometries in solid propulsion systems. Through the URSA program, he looks forward to guiding students who share his passions, and helping them grow in these fields alongside him. Description of Possible Projects: Project 1: Satellite Orbit and Attitude Control System Design Using MATLAB Project Description: This project provides a comprehensive introduction to both orbital mechanics and satellite attitude control. Students will use MATLAB to model and analyze satellite orbits and reference frames, then extend their knowledge by designing and simulating an adaptive attitude control system for maintaining satellite orientation. The project will cover the fundamentals of orbital mechanics, spacecraft orientation, and control systems, culminating in the development of a robust control model that can adapt to various disturbances. Desired Skills: Required: Basic knowledge of MATLAB, interest in orbital mechanics, basic understanding of control systems Optional: Background in Aerospace Engineering, Electrical Engineering, or Physics Timeline & Deliverables: Weeks 1-3: Introduction to MATLAB, basic orbital mechanics, coordinate systems, and reference frames. Weeks 4-7: Modeling and simulating satellite orbits in MATLAB, including both circular and elliptical orbits. Introduction to satellite attitude dynamics and control system fundamentals. Weeks 8-10: Development of a basic PID control system using MATLAB, followed by the implementation of an adaptive control algorithm to maintain satellite orientation under simulated disturbances. Weeks 11-12: Final project report and presentation, summarizing orbital models, control system design, and system performance analysis. Project 2: Design and Performance Analysis of Grain Geometries in Solid Rocket Propulsion Project Description: This project focuses on understanding the impact of grain geometries on the performance of solid rocket propulsion systems. Students will explore different grain shapes, analyze how they affect thrust and burn time, and create models to predict performance using basic simulation tools. Desired Skills: Required: Basic understanding of propulsion systems, familiarity with simulation tools or software Optional: Background in Mechanical or Aerospace Engineering Timeline & Deliverables: Weeks 1-2: Introduction to solid rocket propulsion and grain geometries. Weeks 3-5: Selection and modelling of different grain geometries, understanding burn rates and thrust curves. Weeks 6-8: Simulation of selected grain geometries and performance analysis. Weeks 9-10: Comparison of simulation results with theoretical predictions. Weeks 11-12: Final project report and presentation on the design and performance analysis. Desired Skills and Time Commitment: Project 1: Required: Basic knowledge of MATLAB, interest in orbital mechanics and control systems, willingness to learn and engage with new concepts Optional: Background in introductory-level Aerospace Engineering, Physics, or related fields; familiarity with basic mathematics and programming concepts Time Commitment: Weekly Commitment: 8-10 hours per week Project Duration: Full semester Overall Expectation: Students should be prepared to commit to consistent weekly work, including modelling, simulation, and report preparation. Project 2: Required: Basic understanding of propulsion systems, familiarity with basic simulation tools or software, interest in rocket propulsion Optional: Background in Mechanical or Aerospace Engineering, experience with CAD software or basic simulations Time Commitment: Weekly Commitment: 8-10 hours per week Project Duration: Full semester Overall Expectation: Students should be prepared for consistent weekly work, including modelling, simulation, and analysis. |
| Shensheng Zhao (Electrical and Computer Engineering)  | Shensheng is a 5th year PhD student and am a Beckman graduate fellow and Mavis Future Faculty fellow. His current research interests focus on biomedical photoacoustic and ultrasound imaging for diagnosis and therapy. He hopes to work with great undergrads to develop novel biomedical techniques that benefit people’s lives. Description of Possible Projects: We offer two exciting projects for students: Project 1: Ultrasound Neuromodulation System Development This project provides an excellent opportunity to delve into the world of biomedical engineering. Students will work with cutting-edge equipment, mastering its operation, and gaining hands-on experience in equipment integration using LABVIEW. Preferably, students should be familiar with electronics devices, like oscilloscopes, RF amplifiers and functional generators. Project 2: Deep learning based Vascular Segmentation Algorithm Development This project is programming oriented. Students will receive access to comprehensive vascular datasets, and they will learn how to develop deep learning models and analyze vascular images. Desired Skills and Time Commitment: Desired Skills: Students are required to have high motivation and a willingness to learn diverse knowledge are essential attributes for project participation. Knowledge of LABVIEW and MATLAB is preferred but not mandatory. Time Commitment: Students should commit to a minimum of 8 hours of weekly lab work. |
| Shiau-Jie Rau (Astronomy)  | Shiau-Jie is a third-year PhD student in the Department of Astronomy, working with Prof. Paul Ricker. Her research interests include binary stars, hydrodynamics simulations, and computational astrophysics. She has previously worked on using hydrodynamics simulation to study Type Ia supernovae. Currently, her research focuses on binary systems that once went through common envelope evolution– which means that the two stars in a binary system share the same envelope. Description of Possible Projects: Influence of different donor-companion mass ratios on common envelope evolution In common envelope evolution, one member of a binary star system engulfs the other, causing the two stars to spiral together. During common envelope evolution, the outer parts of the donor star will be ejected, forming a dense cloud of gas concentrated about the equatorial plane. Different companion-donor mass ratios could affect the amount of gas that is ejected. In this project, we will use hydrodynamical approaches to simulate gas interactions in a common-envelope binary. This project consists of three sub-projects, each of which can be taken on by one or more students. Project 1: Using the stellar evolution code MESA (Modules for Experiments in Stellar Astrophysics) to produce donor star models. We will start by producing red giant star models with different masses. Project 2: Using 3D hydrodynamical simulation codes SPHARG and FLASH4 to perform common envelope simulations. Project 3: Using the yt package under Python to analyze the data from FLASH4 simulations. Project 1 and Project 2 will include choosing parameters and running programs on supercomputers. For Project 3, you will write Python scripts using the yt package to reduce the data from the simulations and make plots and movies. Desired Skills and Time Commitment: Students will be best prepared for this work if they have knowledge from ASTR 210 (Introduction to Astrophysics) and ASTR 310 (Computing in Astronomy), or have relevant experience. I am willing to work with students who have less experience, but the expectations for progress in one semester should be tempered in that case. I recommend students be enrolled in or have taken the PHYS 2xx and MATH 2xx sequences. The main requirement is that students be unafraid of programming and be willing to take the initiative to learn computer skills they may not already have. They should expect to work 2-5 hours per week on the project. |
| Shitao Shi (Computational Science and Engineering)  | Shitao is a dedicated fourth-year Ph.D. student in Structural Engineering with dual concentrations in Computational Science & Engineering and Data Science and Engineering within the Civil and Environmental Engineering department. His research aims to mitigate building damage from earthquakes and windstorms through the application of natural hazard engineering principles, innovative computational methods, and data science techniques. Although his research is highly interdisciplinary and spans several areas, his mentees will focus on one or two key aspects that align with their interests. High-achieving mentees may have opportunities to advance the research further or apply for research awards in the following year. Description of Possible Projects: Project 1: Applications of Artificial Intelligence and Machine Learning in Natural Hazards Engineering This project focuses on exploring and implementing artificial intelligence (AI) and machine learning (ML) techniques within the field of Natural Hazards Engineering. The first phase of the project involves reviewing cutting-edge research papers to understand current theories and applications of AI and ML in natural hazard scenarios. In the second phase, mentees will gain hands-on experience with AI-enabled tools and develop machine learning models to enhance earthquake engineering workflows. Timeline: - Weeks 1-3: Review of current literature on AI and ML applications in Natural Hazards Engineering. This includes identifying key research papers and understanding theoretical foundations. - Weeks 4-6: Learning a new AI-enabled tool designed to assist with regional-scale simulations. This phase involves both learning the tool and integrating it into existing workflows. - Weeks 7-9: Development and training of a neural network model to automatically classify and identify soft-story buildings. This model will be aimed at improving the accuracy of earthquake impact simulations. - Weeks 10-12: Evaluation and refinement of the neural network model, followed by integration into earthquake engineering workflows. Preparation of a comprehensive report documenting findings, methodologies, and results. Expected Deliverables: - A detailed literature review report highlighting advancements in AI and ML applications in natural hazards engineering. - A trained neural network model with documented performance metrics and its application in identifying soft-story buildings. - A final project presentation poster summarizing methodologies, results, and implications for future research. Project 2: Bayesian Calibration of Steel Material Model Parameters for Finite Element Simulations This project focuses on applying Bayesian inference techniques to calibrate parameters of the steel material model within OpenSees. The calibration will utilize experimental stress-strain data from cyclic tests on steel test coupons to enhance the accuracy of stress response predictions under different loading conditions. Timeline: 1. Weeks 1-2: Literature review on Bayesian calibration methods and finite element analysis simulations for material modeling. 2. Weeks 3-4: Data preprocessing from experimental stress-strain tests. 3. Weeks 5-6: Implementation of Bayesian inference techniques for parameter estimation in the steel material model. 4. Weeks 7-9: Validation and refinement of the calibrated model through simulations. 5. Weeks 10-12: Analysis of results and preparation of a final report. Expected Deliverables: - Calibrated steel material model parameters and associated data. - Simulation results demonstrating the model’s accuracy. - Final report detailing methodology, results, and conclusions. Desired Skills and Time Commitment: General Requirement: Proficient in Python, with knowledge of NumPy, Pandas, and other data processing libraries. Project 1 Optional Requirement: - Experience with machine learning frameworks (e.g., TensorFlow, PyTorch) - Background in structural engineering, familiarity with natural hazards engineering and regional-scale simulations. Project 2 Optional Requirement: - Background in statistics and related data science majors, familiarity with knowledge of Bayesian inference. - Background in structural engineering or other engineering majors, familiarity with finite element analysis (FEA) software (preferably OpenSees). Time Commitment: Approximately 10 hours per week. |
| Yueze Liu (Electrical and Computer Engineering)  | Yueze is a second-year Masters ECE student. He earned two bachelor's here, physics and computer engineering in 3 years. He says "I LOVE science and everything related to it." His experience ranges from Reinforcement learning/Attention model application, operating systems, IOT microcontrollers/PCBs, and electronic and photonic crystal fabrication. He has 3 papers under bio-ML, but these days he likes IOT and hardware more. Description of Possible Projects: Project 1 - Chat with waifu in real time! TTS synthesizing with sovits and fine-tuning character response and voice generation I made a talking chatbot that uses API's, you can try it out using the static server here: https://ec5a-153-33-112-212.ngrok-free.app The problem is that the quality of voice is still able to be improved, and the response time of deployment can be significantly reduced. This project really just needs one to be able to read code and understand it fast honestly. I went into this project not knowing what an html or any TTS algorithms is and the progress so far has been just over 3 weeks lmao. You will learn a lot though, how TTS works, how servers/IPs work, how to debug/refactor code efficiently (trust me industry will love you once they see your bug log) and hardware stuff you never seen before! Try this out if you want a job later - trust me, they'll want you Required - curious, energetic Skills required - none Skills required to learn while training- python, how to use GPT, read XD Deliverables - improved TTS quality, Server migration/upgrade, Pipeline for massive TTS model production Timeline - improved TTS quality - a week; Server migration - 10 hours; Pipeline - 10 hours Project 2 - Optical Metasuface Mask with high transmission coefficients - towards an Optical Computing platform This project is a real beast. We make this we can patent up and go face NVidia. The overarching goal is that to make an optical computer, one has to have a compact, low loss environment for optical signals. (Optical computers are 100000000% better than classical computers btw, if we can make it work). Recently most other universities has started research on this, and are receiving a lot of funding, but our university don't. We will investigate ways of using electronically induced bricks to alter refractive indices. What you'll learn: How to read papers, FAST. And be able to make judgements, extract key contents, skip over unnecessary details, and how to spend/save money! (Seriously, I have 65000 in funding for this project, if you like to try to do the ""budgeting"" please be my guest). Timeline - literally, if we're lucky it might take a week to brainstorm a setup and a week to fab. If we're unlucky, it would require us the amount of time that our hair would've stayed on our head /6 (probably) to get a model that works on paper. Desired Skills and Time Commitment: Desired skills - Senior software engineer/Senior hardware engineer Actual required skills: Passion for life Skills used for projects: python finding out what to use for projects learning how to use them |