Projects

llm-eval-pipeline

Automated LLM output evaluation: deterministic scoring checks and rolling-window drift/regression detection across model runs.

vector-db-benchmark

Benchmarks vector search backends (pgvector, Qdrant, FAISS) on query latency and recall@k against a brute-force ground truth.

llm-dataset-curation

A distributed pipeline that curates a raw text corpus (Project Gutenberg) for LLM pretraining/fine-tuning: exact + near-dedup, quality filtering, chunking, and export.

StepTracker

Simple way to track and visualize your daily step count