
Product engineer // AI systems
I define the problem, write the requirements, and build the systems behind them. I lead AI across ServiceOps at Motadata.
// Selected work
AI systems I've specified and built, plus earlier web work.
Motadata ServiceOps, in production
ServiceOps' agentic AI suite. IT teams deploy pre-built AI agents and manager-orchestrated AI teams with no code, and reach them through the in-portal Ask AI assistant.
- Manager agent
- Orchestrates the team
- AI agents
- Pre-built, no code
- Knowledge
- RAG on the customer's own data
- PII guard
- Masks input and output
- Test and preview
- Before every go-live
Agentic AI / multi-agent orchestration / RAG / AI governance
LLM gateway
An OpenAI-compatible gateway that puts one endpoint in front of multiple model providers, with a React ops console and one-command Docker deployment.
- Budgets
- Per team, usage metered
- Rate limits
- Enforced atomically
- Semantic cache
- Serves repeat questions
- Routing
- Easy requests to cheaper models
- Failover
- Timeout, retry, next provider
Python / PostgreSQL / pgvector / Redis / React / Docker
Explainable candidate matching
A hybrid RAG engine that ranks candidates against job descriptions and explains every result.
- Semantic search
- ChromaDB
- Keyword search
- BM25
- Fusion
- Reciprocal Rank Fusion
- Must-haves
- Checked before scoring
- Score
- 0 to 100, with a reason
Python / ChromaDB / OpenAI embeddings / BM25
Realtime chat application
A real-time chat system built with Django Channels and HTMX.
- Messaging
- Personal and group
- Files
- Shared in chat
- Presence
- Live
- Transport
- Channels over Redis
Python / Django / HTMX / Redis / ASGI
- Catalog
- Category filtering
- Cart
- Session-aware
- Checkout
- Order workflow
- Accounts
- Persistent login
Python / Django / HTML / CSS
// How I work
I define the problem, write the requirements, and build the systems behind them.
End-to-end ownership
From problem discovery and PRDs to architecture, implementation, tests, and launch.
Applied AI engineering
Agentic systems, RAG, LLM gateways, and evaluation against baselines.
Production judgement
I design for cost control, failover, governance, and data privacy from day one.