Lab — Yegor Gaidar

WebLLM In-Browser Chat

On-device LLM running entirely on your GPU via WebGPU. Multiple model sizes (0.5B to 1.5B), streaming responses, zero API calls. No budget for API hosting, so I shipped the model to the user's GPU instead.

WebGPU LLM WASM Qwen2.5

Standalone demo →

Rust/WASM Fraud Detection Benchmark

Interactive benchmark comparing JavaScript vs Rust/WASM fraud detection scoring. Runs 10K–250K synthetic transactions through a 7-rule pipeline — velocity checks, amount anomalies, geo-impossible-travel, device fingerprint reputation, and merchant risk scoring.

Rust WASM TypedArrays Benchmark

Run the benchmark → · Read the write-up

ArchDogma

Honest static analysis for Python — catch circular imports, god modules, and tight coupling before they become postmortems. Open-source core on PyPI plus a paid Pro Kit.

Python AST Static Analysis PyPI

Project page → · GitHub