Founding Engineer · Digital Wisdom (WonderCal) ·San Francisco
I build AI-native products end to end, from agent tooling to the production infrastructure behind it.
Highlights from what I've been shipping lately.
Digital Wisdom (WonderCal)
18 MCP tools that let LLM agents book and manage meetings (retry/backoff, token refresh, HMAC-signed OAuth callbacks), with a Claude API fallback validated by the existing parser.
Digital Wisdom (WonderCal)
Built and hardened integrations across Google Calendar, Microsoft Graph, and iCloud CalDAV over OAuth 2.0, diagnosing sync failures on large calendars, permission scopes, and timezone boundaries.
University at Buffalo
Semantic chunking and Pinecone vector retrieval extracting structured fields from scanned forms: 5,000+ records at 90%+ accuracy, cutting ~25 hours a week of manual review.
Selected independent work in agentic AI and backend systems.
Four agents coordinating 44 tools across 6 MCP servers, with structured tool-failure handling, per-call timeouts, and health-based routing. Provisioned with Terraform on AWS; Prometheus monitoring.
Python · OpenAI Agents SDK · MCP · LangGraph · Terraform · AWS
Multi-tenant starter with Stripe subscription billing, authentication, role-based access control, and tenant data isolation.
FastAPI · PostgreSQL · Stripe · Docker · AWS
Asynchronous code review pipeline built on FastAPI, Redis, and Celery, with LLM-backed analysis via OpenAI and LangChain.
FastAPI · Redis · Celery · OpenAI API · LangChain
I'm open to conversations about engineering roles, agent infrastructure, and interesting problems in AI-native products.
$ reading: Inference Engineering · Philip Kiely (Baseten)