Services
A system isn't finished when it works on your laptop. It's finished when it runs on real data, in front of real users, with someone accountable for the output. Everything below is priced and scoped to reach that point.
What I do
Most engagements draw on two or three of these at once — a model is only as good as the pipeline feeding it, and an agent is only as useful as the system it can actually reach into.
Multi‑agent pipelines that produce evidence‑backed, auditable output rather than plausible prose. Custom agent development kits, MCP and agent‑to‑agent architectures, retrieval, fine‑tuning, and evaluation harnesses.
Typical problems: document generation that has to stay internally consistent across a whole corpus; extraction from messy records; workflows where a wrong answer is worse than no answer; inference costs that don't survive contact with volume.
Prediction systems that run every day rather than in a notebook. Distributed training and deployment, drift and bias monitoring, and the integration work connecting a model to the systems people already use.
Typical problems: a model that performs in validation and not in the field; no monitoring, so nobody knows when it degrades; predictions nobody acts on because they arrive somewhere no one is looking.
The plumbing that decides whether any of it works. EHR and FHIR integration, lab instruments, LIS, CRM, warehouse. ETL and feature pipelines from prototype to production, with monitoring that holds up under audit.
Typical problems: data that exists but can't be joined; manual reconciliation nobody has time for; integrations that break silently; reporting the business argues with instead of using.
Claims forecasting, financial risk modeling, cohort and outcomes analysis, supply chain vulnerability. Numbers a leadership team can act on and defend.
Typical problems: forecasts that miss badly enough to be ignored; risk scores without an intervention attached; analysis that can't be explained to the person who has to sign off on it.
How to work with me
A working session and a short investigation: the decision the system has to support, the data that actually exists, and what good has to look like. You leave with a written scope, an architecture, and a fixed price for the build — whether or not you do it with me. Most engagements start here, and some usefully end here.
The work itself, in short cycles that each end with something running. You see the real system on your real data instead of a deck describing one. Scope changes get priced before they get built, and the engagement ends with deployment, monitoring, documentation, and your team trained to own it.
Architecture review, technical due diligence, hiring, and a second opinion on build‑versus‑buy. For teams with engineers who need direction more than hands. Some advisory relationships expand into hands‑on delivery; that's a separate conversation and a separate rate.
When a project needs more than one person, I bring in a bench I've worked with for years — four data scientists and two project managers — so you scale without being handed a stranger. For work needing a full delivery organization behind it, I partner with Alithi Management Consulting.
How an engagement runs
Define the decision, audit the data, agree on what good looks like. Ends with a written scope and a fixed price.
Short cycles, each ending in something that runs. Real data, not a sandbox. Scope changes priced before they're built.
Deployed into your stack with monitoring, documentation, and your team trained to run it. The goal is to make myself unnecessary, not retained.
Start here
I take on a small number of engagements at a time, so I can be straight with you about whether yours is one I should take. Bring me the one that stalled.