I am Sirui Li, an avid researcher with a profound interest in the realms of automated theorem proving, logic, and artificial intelligence, alongside my exploration of various pure mathematics domains. My academic pursuit is centered on deciphering the delicate boundary where machine capabilities intersect with human intellect, particularly in mathematics. Through my studies and research at New York University and collaborations with leading institutions, I am dedicated to advancing our understanding of how AI can enhance and transform the field of mathematics, driving forward both theoretical and practical applications. Join me as I navigate this fascinating intersection of technology and human knowledge.
🔥 News
- 2024.08: 🎉🎉 Started a position as technical consultant for 10 accounting firm, developed an AI agent for training new coming employees. Used tech stack: RAG, Graph RAG, FastGPT, Dify, Ollama, etc
🎖 Honors and Awards
- 2024.06: Among only 20 participants selected from all over China, was accepted to AI4Math Research Program at PKU, awarded a prestigious prize worth 160,000 RMB (22,200 USD)
- 2024.04: With $6K award, Ray is accepted to NYU Abu Dhabi Undergraduate Summer Research Assistantship.
- 2023.12: With $12K award, Ray is accepted to NYU Abu Dhabi Undergraduate Research Assistantship.
- 2023.11 Accepted to Microsoft Startup Program (1% acceptance rate) with $18K worthy prize; Accepted to NYU startup bootcamp.
- 2023.07 Dean’s List for the 2022-2023 academic year
- 2023.02 “Best First Time Hacker” with $200 prize in the Hackathon at New York University
🔬 Research Experience
- 2024.07 - Present, Automated Theorem Proving & Formal Verification, Peking University
Research Assistant Under Prof. Zaiwen Wen
- formalization of the ADMM algorithm in convex analysis using the Lean language https://lnkd.in/etcYRYBP
- Contributed to the Reaslab IDE, including recording introductory videos and testing the system (both group and individual tests) https://alpha.reaslab.io/
-
2024.01 - 2024.07, Investigating Transformer-based Model on non-stationary time series data
Paid Research Assistant Under Prof. Djellel Difallah
- Developed auto-experiment scripts, boosting research productivity by over 150\% and facilitating extensive model performance evaluation Github Repo.
- Visualized experiment data with interactive callback real time calculation interface (Github Repo)
- Discovered precision improvements in stationary data forecasting with Autoformer when input vector length exceeded data period length, and the model’s enhanced performance with higher data frequencies.
- Discovered the relationship between prediction precision with the input vector length, label length, prediction length, and data period length.
- Analyzed predictive patterns in both stationary and non-stationary data, highlighting Autoformer’s effectiveness in non-stationary contexts with its auto-correlation feature, surpassing standard transformers.
- 2023.01 - 2023.05, Research Assistant Under Prof. Megan Morrison
- Investigating when eigenfunctions for ODEs can be generated from invariant manifolds
- Helped on research about utilizing Koopman operator theory to linearize non-linear ordinary differential equations.
- Helped professor wrote a Python program to predict eigenfunctions.
- Engaged in self-directed learning to understand Ordinary Differential Equations (ODEs), focusing on solving a group of ODE equations independently.
📖 Educations
- 2024.01 - 2024.06 (now), In Progress: Visiting AI/ML Research Assistant under Professor Djellel Difallah, New York Universty Abu Dhabi, Abu Dhabi, United Arab Emirates.
- 2022.09 - 2026.07 (now), In Progress: Bachelor of Arts in Math (honor) and Computer Science, New York Universty, NYC.
- 2022.06 - 2026.08, Summer School: Math 54 -Linear Algebra And Differential Equation: A; CS 61B. Data Structures: A- , University of California Berkeley, Berkeley, California.
🏫 Courses Taught & Taken
I will only list the courses related to Computer Science and Math.
-
2024 Fall, Will Take:
- MATH-GA 2430 (Graduate) Real Variables I, Tristan Buckmaster.
- MATH-UA - 348 Honors Algebra I, Sylvain Cappell.
- CSCI-UA.0520/0521 Undergraduate Research, Thomas Wies. Focus on Formal Verification.
- PHIL-UA 73 - 310 Set Theory, James Walsh.
- VIP-UY 300X Vertically Integrated Projects, Siddharth Garg
-
2024 Spring, Took:
- MATH-UH - 2010 Ordinary Differential Equation, Arman Mimar.
- MATH-UH - 3612 Differential Geometry, Hisham Sati.
- MATH-UH - 3413 Numerical Methods, Francesco Paparella.
- MATH-UH - 3610 Complex Analysis, Ilya Spitkovsky
- 2023 Fall, Taught: MATH-UA 121 Calculus I, Teaching Assistant
- 2023 Fall, Took:
- MATH-UA 328 Honors Analysis I, taught by Jalal Shatah.
- CSCI-UA 201 Computer Systems Organization, taught by Benjamin Goldberg.
💻 Internships
-
2023.06 - 2023.09, 300K.xyz (First Tier company in DEX) Software Development Engineer Internship at Beijing, China
- Designed and Developed Real-Time Data Processing Systems: Architected a live signal querying system, enhancing the data reading speed by 40% through a bespoke algorithm. This system also allowed for real-time prediction capabilities with time series machine learning, integrating features for error-handling and robustness, ensuring seamless tracking of the latest trading data and predictions. Developed an interface connecting Python with C++ to replace existing Python code, achieving a 17x speed improvement.
- Optimized Data Querying and Management: Introduced an auto-configuration generation process paired with a set of RESTful APIs. These APIs, coupled with a specialized volume-check algorithm, expedited the query speed by 80%. Also established data validation stages that tackled the issue of discontinuous data, ensuring chronological continuity and accurate predictions.
- Developed Versatile Cryptocurrency SDK: Pioneered a comprehensive Python SDK library for cryptocurrency trading, covering a range of functionalities from coin transactions to liquidity management. The SDK, structured with utilities such as order handling, position management, and network operations, serves as a one-stop solution for cryptocurrency operations.
👨💻 Open Source Projects
Filling gaps in chaotic time series
Sirui (Ray) Li
-
Advanced Mathematical Techniques in Chaotic Systems: Developed a software tool leveraging Takens’ Theorem for dynamic reconstruction from incomplete time series, specifically addressing the challenges of filling gaps in chaotic data. This method ensures the continuity and dynamical integrity of the reconstructed data, demonstrating a sophisticated application of theoretical mathematics in practical data analysis scenarios.
-
Software Architecture and Optimization Strategies: Engineered a modular software architecture incorporating various strategies for calculating dynamics, estimating vector fields, and determining minimal distances. This design enhances the software’s flexibility and scalability, making it adaptable for different chaotic systems and data types.
-
Proven Effectiveness and Innovative Problem-Solving: Demonstrated the effectiveness of the gap-filling software using synthetic data generated from the Lorenz system, achieving significant improvements in data consistency and error minimization. This project not only underscores my ability to apply complex mathematical concepts but also highlights my skill in optimizing algorithms for robust, real-world applications in scientific research.
Sirui (Ray) Li, Ge Jin, Davyd Viniarskyi, Monique Armelle Dacanay
- Developed “Explain”, a Chrome extension leveraging GPT-3.5 and the youtube_transcript_api to extract and analyze YouTube video transcripts.
- Implemented a binary search algorithm for pinpointing user queries within videos, integrating OpenAI’s API for accurate, context-based responses.
- Engineered a robust backend using RESTful API on AWS EC2, complemented by a dynamic JavaScript frontend for an intuitive user experience.
Previous News
- 2024.06: Decided next semester’s schedule: Graduate Level Real Variable, Honors Algebra, Principle of Programming Language, Operating Systems
- 2024.05: 🎉🎉 Final Grade: Differential Geometry get an A; Complex Analysis get an A.
- 2024.3: Started contributing to the SymPy Library
- 2024.2: 24 Spring classes taking: Differential Geometry, Complex Analysis, Numerical Analysis, ODE.
- 2023.12: 🎉🎉 With $6K award, Ray is accepted to NYU Abu Dhabi Undergraduate Research Assistantship.
- 2023.12: 🎉🎉 Get an A for MATH-UA 328 Honors Analysis I taught by Professor Jalal Shatah. Only 4 students got an A out of more than 70 enrolled students.
- 2023.12: 🎉🎉 Semester GPA: 4.0