I work with engineering teams on a project or advisory basis. I don’t do vague strategy decks. Every engagement ends with something built, decided, or unblocked.
I take on a small number of projects at a time. If the problem is specific and the timeline is real, read on.
Production AI System Build
A focused 4–8 week engagement to take a defined AI capability from prototype to a production-ready system. Not a polished demo — a system that runs on real data, has monitoring, and keeps working after the engagement ends.
What’s included:
- Architecture scoping and decision review at the start
- Hands-on build: inference pipelines, model serving, orchestration, deployment on AWS
- Documentation of how the system works and how to operate it
- Handoff session with your team
Who it’s for: Engineering teams with a POC they can’t get to production, or teams starting a new AI initiative and wanting to build it right the first time. Especially useful when your team’s core strength is product or backend, not ML infrastructure.
Architecture Review
A 1–2 week technical review of your AI or data platform architecture. You walk away with a written assessment of what’s working, what will break at scale, and a prioritized set of recommendations tied to your actual roadmap.
What’s included:
- Review of your current stack, pipelines, and AI system design
- Written assessment: strengths, risks, gaps
- Prioritized recommendations
- One working session with your technical team to go through findings
Who it’s for: Teams about to make a significant investment in AI infrastructure and want a credible second opinion before committing. Also useful for teams inheriting a system they don’t fully trust.
MCP and Agent Integration
A project engagement to connect your AI system to internal data and tools using the Model Context Protocol. This includes building MCP servers, wiring agents to internal data sources, and building lightweight internal copilots that work on production data.
What’s included:
- Assessment of what data and tooling your AI system needs access to
- MCP server design and implementation
- Integration with your LLM interface (Claude, OpenAI, etc.)
- Documentation and team walkthrough
Who it’s for: Teams that have deployed an LLM interface — internal chat, copilot, assistant — but it can’t access the data it needs to be actually useful. Also for teams exploring agent architectures that need to connect to real internal systems.
Not sure which fits?
If you have a real problem and you’re not sure whether any of these apply, send me a short description of what you’re building and where you’re stuck. I’ll tell you honestly whether I can help.