Hello, I am Huanfu Li
I focus on applying machine learning and backend technologies to build efficient, intelligent solutions.
Professional Background
I am a Computer and Information Technology student at Purdue University, with a minor in Political Science. I am passionate about AI/ML and dedicated to applying cutting-edge technology to solve real-world problems.
During my internship at Chinasoft International, I developed a time series forecasting model that successfully reduced model loss by 90% and achieved a 30% return on investment. This experience gave me deep insights into machine learning applications in the financial sector.
I focus on building efficient, intelligent solutions, excelling at transforming complex algorithms into practical systems. Whether in backend development or machine learning model deployment, I pursue code elegance and system scalability.
Beyond the Code
Beyond technical work, I have a strong interest in political science, which allows me to think about problems from multiple perspectives and understand the impact of technology on society.
I believe my interdisciplinary knowledge background helps me consider more factors in technical decisions, not just technical feasibility, but also social value and ethical implications.
In my spare time, I enjoy exploring new technology trends, participating in open source projects, and sharing my learning insights and technical perspectives through blogging.
Experience & Research
Backend Developer & Machine Learning Engineer Intern
Chinasoft International Limited
- •Delivered a time series forecasting model using Python (LSTM integrated with Multi-Head Attention and Residual Networks) achieving up to 30% overall investment returns
- •Reduced model average loss by 90% through manual tuning and hyperparameter search using Optuna
- •Integrated early stopping techniques to prevent overfitting and reduced training costs by 40%
- •Improved test coverage by 25% through AI-assisted unit testing generation and code auditing
Undergraduate Data Science Researcher
The Datamine Project – Purdue University
- •Built a manager coaching tool based on ChatGPT 3.5 Turbo LLM for HUMN Capital to analyze employee engagement and predict turnover risk
- •Used Whisper API to process video content, converting unstructured data to structured text for sentiment analysis
- •Worked as part of an Agile team using Kanban and Tableau to manage software projects and allocate work
- •Utilized Git for efficient version control and seamless collaboration on team repositories
Featured Projects
Project Iris - Always on XR AI Assistant (WIP)
Extended Reality application for Meta Quest 3 featuring an always-on AI assistant using Gemini Live API with Unity C# frontend and Java Android backend. This project is a work in progress, please check the button for video demo.
- •Unity C# frontend with immersive XR interface design
- •Java Android backend with real-time API integration
- •Gemini Live API for natural conversation capabilities
Bitcoin Price Prediction Model
Time series forecasting model based on Transformer architecture, integrating various financial technical indicators, open-sourced under MIT license to aid quantitative trading strategy development
- •Reduced model average loss by 90%
- •Integrated early stopping to prevent overfitting
- •Introduced feature engineering to enhance model performance
AI Code Debug Assistant
Developed a standalone full-stack debugging assistant based on DeepSeek R1 7B LLM, reducing MTTR by 28%
- •Implemented strategy design pattern for database interfaces
- •Designed bidirectional duplex server for concurrent connections
- •Used FileStream for direct source file manipulation
- •Deployed MySQL database for data management and user Auth
Tech Stack
AI & Machine Learning
Backend & Database
Frontend & Others
Tools & Platforms
Education
Purdue University
B.S. in Computer and Information Technology
Minor in Political Science
08/2022 – 05/2026
GPA: 3.81/4.00
Dean's List & Semester Honors, Fall 2022 – Spring 2025
Hobbies & Interests
Photography
I enjoy documentary and landscape photography, as well as film photography. I aim to capture contemporary life through the lens, reflecting the spirit of our times.
3D Printing
Designing and fabricating prototypes using 3D modeling and additive manufacturing techniques. The ground effect vehicle model in the picture is designed and printed by me, aiming to explore efficient low-altitude flight technology.
Racing Cars
Passionate about automotive engineering and competitive racing, constantly seeking to improve driving skills and vehicle performance.
My Technical Vision
I believe software is the architecture of the 21st century. Our future social systems—from public services to financial compliance—will be built on code. However, we currently build this critical infrastructure using 'manual workshop' methods.
This bottleneck doesn't just slow innovation; it makes building the transparent, fair, and efficient systems we need to replace outdated bureaucracies almost impossible. My goal as a Responsible Systems Architect is to build the next generation of AI Governance Systems.
But this vision faces a critical barrier: Trustworthiness. How can we verify that a system managing millions of lines of code is fair, secure, and aligned with human rules?
My mission is twofold: First, to develop the foundation: an 'AI for Code' platform that acts as an 'AI Auditor.' This isn't just about speed; it's about Trust-by-Design. By leveraging Formal Verification, this AI will automatically validate that code is secure, reliable, and compliant with ethical rules before it's deployed.
Second, to apply this foundation to build Reliable & Human-Centered AI systems that can accurately translate human policies (like law or regulations) into auditable, executable logic.
I am convinced that without a robust platform to prove their reliability, such complex systems should not be built at all—they would be too complex to manage and too opaque to trust.