Engagements built around your problem, not my stack.
Three ways to work with me. Pick the one that matches the shape of your problem and your timeline.
Sprint
One narrow problem, one shipped solution. Best for: 'we need a working RAG/chatbot/forecast model and we need it before next quarter.'
- ✓Discovery & written brief
- ✓Working prototype on real data
- ✓Eval pipeline + observability
- ✓Documented handover to your team
Embedded
I work as your part-time AI engineer. 2–3 days a week, deep in your codebase, in your standups, on your Slack.
- ✓20–30 hrs / week
- ✓Architecture, code, hiring help
- ✓Direct access on Slack/Linear
- ✓Weekly written status
Advisory
Ongoing technical advisor. Code reviews, architecture calls, vendor selection, hiring panels.
- ✓Bi-weekly 60-min calls
- ✓Async Slack / async review
- ✓Architecture decision docs
- ✓Hiring & vendor support
Everything I build.
Not every project needs all of these. The discovery call narrows it down.
AI chatbots
Customer support, internal helpdesk, knowledge concierge.
RAG systems
Search & chat grounded in your contracts, docs, code, tickets.
AI agents
Multi-step workflow agents that book, draft, classify, escalate.
LLM apps
Custom copilots, assistants and chat experiences with eval & streaming.
ML model dev
Forecasting, scoring, ranking, classification — trained on your data.
MLOps
Deployment, eval pipelines, monitoring, cost & retrieval-quality dashboards.
AI consulting
Architecture reviews, vendor selection, build-vs-buy, hiring panels.
Fractional AI lead
Embedded part-time or full-time. Ship code, hire, set the roadmap.
How fast can you start?01 / 06
I usually have one slot opening per month. The earliest is typically 2–3 weeks out — but I always reply within 24 hours so we'll know quickly.
Do you work with non-technical founders?02 / 06
Most of my clients are non-technical or technical-but-not-AI. I write the brief in plain English and translate it into engineering. That's half the job.
Can you white-label for my agency?03 / 06
Yes — about a third of my work is sub-contracted through agencies. I'll show up under your brand, your client never sees my name unless you want them to.
Do you take equity?04 / 06
For early-stage companies, partial-equity engagements are possible after the first paid sprint. I won't do equity-only.
What stack do you use?05 / 06
Whatever fits. Usually: Python, FastAPI, Postgres, Pinecone or pgvector, OpenAI/Anthropic/open models, LangGraph or hand-rolled orchestration, Modal/Render/AWS for serving. Stack follows the problem.
Do you sign NDAs?06 / 06
Always, before any code or data is shared. The discovery call doesn't require one.
Got a problem AI might solve?
30 minutes. Free. You leave with a clear yes/no on whether to build, and a one-pager you can forward to your team.