Hi, I'm @reese8272 — an AI engineer from the Eastern Panhandle of WV. I build and ship agentic AI systems: I run a live production SaaS (autoclip.studio) and build multi-agent LangGraph systems and MCP servers professionally at Cognizant. Outside of code I'm into bowling, pole vault, Rubik's cubes, and relentlessly tuning my own learning habits.
- Languages: Python, SQL, TypeScript, C
- AI / Agents: LangGraph, LangChain, MCP servers, RAG, pgvector, Anthropic & OpenAI APIs, WhisperX
- Backend: FastAPI, Celery, PostgreSQL, Redis, REST
- Frontend: React, TypeScript
- Infra / Tooling: Docker, CI/CD, AWS, Azure, Stripe, Git/GitHub, Prometheus, Sentry, OpenTelemetry
-
CreatorClip — autoclip.studio — ✅ live production SaaS I built end-to-end. Turns long-form YouTube content into short-form clips, learning each creator's style via a custom "DNA" module.
- Architecture: React/TypeScript frontend + FastAPI backend; async Celery pipeline (WhisperX transcription → Voyage AI + pgvector semantic embeddings → ffmpeg rendering); YouTube OAuth; Stripe billing with a credit ledger; per-creator PostgreSQL row-level security; staging/prod Docker with CI auto-deploy; Prometheus/Sentry/OpenTelemetry observability; ~79% test coverage.
- Repo: github.com/reese8272/creatorclip
-
CFO Agent — 🚧 agentic personal-finance assistant (in active development).
- Architecture: an 11-node LangGraph reasoning agent (retrieval → analysis → specialist → coach → tracker → alert → synthesis → persist) on a FastAPI backend, with PostgreSQL persistence, Redis caching, Anthropic prompt caching, a Fernet-encrypted data "vault," CSV/OFX import, and a financial-scenario engine. Architected multi-tenant.
- Backed by: 159 tests plus an evaluation harness.
Earlier work: Password Manager (Fernet-encrypted local vault + SQLite) · MNIST CNN (convolutional net built and trained from scratch in TensorFlow).
- Email: reesepludwick@gmail.com
- LinkedIn: Reese Ludwick
I can solve a Rubik's cube in about 10 seconds!