Data Science Engineering · Backend & ML Systems
LINKEDIN · EMAIL · GITHUB
I'm a 3rd-year Data Science Engineering student who mostly builds backend systems. I'm drawn to the infrastructure underneath ML and data work, to experiments that are pinned down before they run, and to security designed so the server holds as little as possible.
BACKEND FastAPI services, job queues, CI, auth
EVALUATION tracing, relevance evaluators, offline benchmarks
EXPERIMENTS seeded environments, preregistered metrics
SECURITY client-side encryption, least-privilege GitHub Apps
01 — LLM TRACING & EVALUATION
Tracing and evaluation for LLM applications. A Python SDK sends spans to a FastAPI ingestion API backed by ClickHouse; a Postgres-backed worker scores sampled LLM spans with TF-IDF and embedding-based relevance evaluators, benchmarked offline on WikiQA. Next.js dashboard.
PYTHON / FASTAPI / CLICKHOUSE / POSTGRESQL / NEXT.JS
repository ↗ live demo ↗
02 — CLIENT-SIDE ENCRYPTION
A file vault with client-side AES-GCM encryption via the Web Crypto API. Files are encrypted in the browser before upload; the server stores ciphertext and wrapped keys, never usable key material.
TYPESCRIPT / NEXT.JS / EXPRESS / PRISMA / POSTGRESQL
repository ↗ live demo ↗
03 — CI ANALYTICS
A GitHub App that syncs GitHub Actions run history and runs six statistical analyzers over it: regressions, slow jobs and steps, retry waste, dependency-install overhead. No LLM in the analysis.
PYTHON / FASTAPI / POSTGRESQL / REACT
repository ↗ live demo ↗
04 — LLM RESEARCH · COLLEGE GROUP PROJECT
A research harness for LLM-vs-LLM negotiation experiments, with seeded environments and preregistered welfare and fairness metrics. In progress; no experimental results yet.
PYTHON / ANTHROPIC API / OPENAI API / SQLITE
repository ↗
ALSO SQL · Alembic · fastembed · SQLite · pytest · Vitest
weekly contributions · 251 total · oct 2025 – oct 2026 · peak 45 in the week of 16 aug · snapshot 2 oct 2026
GitHub achievement · Pull Shark ×2



