🧠 1:1 AI Research Mentorship (Summer/Fall 2024)

• Mentor: Edoardo Contente (Stanford-affiliated researcher)

• Duration: 10 weeks (1:1 mentorship) 

• Title: Evaluating Autonomous Vehicle Safety: A Comparative Analysis of ADS and ADAS Incident Data Using Machine Learning Models

• Summary: In this project, I compared the safety performance of Advanced Driver Assistance Systems (ADAS) and Automated Driving Systems (ADS) using machine learning models. By training algorithms on real crash data from the National Highway Traffic Safety Administration (NHTSA), I predicted injury severity based on factors like weather, road conditions, and vehicle movement. Random Forest was the most accurate model, achieving up to 95.98% accuracy. This research highlights how AI can help improve road safety and guide future development of autonomous vehicles.

• Python

• Scikit-learn

• Random Forest, Decision Tree

• Data Preprocessing

• Model Evaluation

I learned how to think like a researcher—formulating questions, building models, and analyzing real-world data. It deepened my interest in using AI to solve meaningful problems.

"Write the paper like water flowing—every sentence should connect smoothly, like a stream with no blockages. Let the logic guide the reader without forcing it."

– Edoardo Contente, Inspirit AI Mentor 

🔍 Inspiration: Distracted Driver project + Trip to San Francisco 

I’ve always been fascinated by how technology can solve real-world problems. As more autonomous vehicles appear on the road, understanding how they perform in critical situations isn’t just a research topic—it’s a question of safety, responsibility, and trust. That’s why I chose to analyze real-world traffic accident data for my 1:1 AI Research Paper.

Inspired by my group’s distracted driving project and a trip to San Francisco that deepened my interest in autonomous vehicles, I wanted to explore how AI can help prevent crashes and improve decision-making on the road. Unlike school research, which often means just collecting information, this was my first time building an actual machine learning model, finding and cleaning a messy dataset, and writing a full-length academic paper from scratch.

It wasn’t easy—the data was chaotic, the preprocessing was exhausting, and I often felt unsure. But I learned to navigate tools like Google Scholar for my literature review, and I relied on ChatGPT to help generate code, which I then refined and debugged. Despite the roadblocks, this experience showed me what real research looks like: uncertain, complex, and deeply rewarding.

It also reminded me that AI isn’t just about models and metrics—it’s about using technology to make life safer and better for everyone.