I am a Software Engineer with 2+ years of experience building full-stack AI platforms, multi-agent systems, and AI-driven data pipelines — shipping production-grade features from schema design and tenant isolation to real-time streaming UIs.
On the backend, I work with Python and FastAPI, building async APIs on PostgreSQL (with pgvector) using SQLAlchemy and Alembic. I design multi-tenant architectures backed by Postgres row-level security, background job processing with Redis-backed queues (arq), and real-time streaming with Server-Sent Events — resumable across dropped connections and idempotent on retry.
I integrate multiple LLM providers (Anthropic, OpenAI, Google, OpenAI-compatible) behind a single adapter layer, and have built tool-calling agent loops with bounded execution, multi-agent delegation, long-term assistant memory, RAG pipelines with crawling, chunking, embedding, and cited sources, and multi-step workflow orchestration. I've also built a tiered web-scraping and structured-extraction pipeline using Playwright and LLM-driven visual browser agents, with semantic search over self-hosted embeddings.
On the frontend, I build with TypeScript, Next.js, React, and Tailwind, creating responsive UIs that consume streaming APIs and manage complex client-side state.
I handle platform concerns like feature flags, usage quotas and cost ledgers, audit trails, and Docker-based deployments fronted by Caddy — and follow disciplined engineering practices: pytest suites against real Postgres and Redis, Playwright end-to-end tests, CI with ruff, mypy, ESLint, and TypeScript checks, feature-branch and PR workflows, and security-first design (Argon2id, encrypted credentials, CSRF and SSRF protection, role-based access control).
