Over the past eight years I have worked on high-traffic products serving 50M+ users, led backend engineers, improved production throughput, and built systems across distributed services, developer infrastructure, and applied AI. Most recently that meant a centralized LLM gateway handling 400K+ requests/day with per-team attribution, quotas and cost governance, and an event-sourced AWS provisioning platform that automated isolated environments for internal product launches.
Most of that work lives in private company repositories. The projects and merged contributions below are public examples of how I approach architecture, performance, and reliability.
Stack: Python (Django, FastAPI) · Go (gRPC, concurrent services) · AWS (IAM, networking, serverless) · LLM infrastructure (LiteLLM, MCP) · PostgreSQL · Redis · Kafka · OpenTelemetry, Prometheus, Grafana, Datadog
- Guardian: A Go microkernel for extensible LLM guardrails, quota enforcement, and policy plugins.
- ContributorCompass: Finds the open-source projects where your skills, growth, and community impact align.
- People Match Engine: Graph- and embedding-based matching of professionals with FastAPI, FAISS, and Redis.
- RateLimiter: Sliding-window and leaky-bucket rate limiting in Go, backed by Redis.
- Django
- Made
makemigrationsshow a symbol for each operation. Every Django developer sees it on each run. - Added a system check for unmatched angle brackets in
path(). - Refactored the serializer's Deserializer functions into classes.
- Made
- LangChainGo
I take on selected engagements in backend architecture, distributed systems, performance and reliability reviews, production LLM infrastructure, and modernizing difficult existing systems.