Hey, I'm Lev. I live in Colorado with my fiancé, where we spend time snowboarding, gravel biking, and exploring the mountains. I’m currently a ML engineer at Honda, where I lead data science and machine learning efforts. I also co-founded an AI-driven startup and studied mathematics at Ohio State.
I'm especially interested in Python, Rust, ML systems, and performance-driven dev tools. You can find my personal projects on this page, and more of my open source work on GitHub.
If you'd like to get in touch, feel free to send me an email.
Co‑founded a tool that lets actuaries interrogate millions of insurance filings (“How has wind‑hail deductibles changed since 2015?”) and get source‑linked answers. A RAG pipeline distilling ~10TB of PDFs into an AI‑powered insurance search engine.
Crash Data API
matrisk.ai/crash-data
100 million police reports, 20 years deep, cleaned and piped into hyper‑local risk scores. Carriers hit the API thousands of times per minute to price telematics and garaging risk.
A Rust‑powered email validator with a thin PyO3 shim. Inspired by Pydantic’s all Python implementation, I re‑wrote the hot path and saw a 100‑1000× speed‑up. Fully RFC 5322/6531 compliant, Unicode‑aware and soon able to ping MX records.
A Rust CLI that audits Python environments, mapping imports → packages, flags bloat and prints disk usage so you can slim Docker images before the next cup of coffee.
Handlefinder
/handlefinder
A fullstack app to search handles across hundreds of social networks. Built with FastAPI, TypeScript, and Node.js. Great for brand checks or OSINT.
April 22, 2025
When we set out to build geospatial risk scores for vehicle crashes at Matrisk AI, we never expected that a side by side look at Vehicle Identification Numbers and crash timelines would hint at possible insurance fraud. But data sometimes surprises you. Below, I’ll walk through how we stumbled upon this discovery, what we found, and why it might matter for anyone insuring vehicles.
