Capabilities · Claims backed by links

What I do, with the proof attached.

A skills list is an assertion. Each capability below links to the documented projects that evidence it. If a capability has no proof behind it, it is not on this page.

Grounded RAG and retrieval

Retrieval systems over difficult real-world documents: slide decks, scans, charts, draft-contaminated and bilingual corpora. Every answer traces to a source; anything unciteable is omitted, not asserted.

See the evidence

LLM evaluation and quality gates

Judge design and calibration, non-circular retrieval metrics, accuracy benchmarking, automated gates that re-run on every change. The measurement harness is often the real deliverable.

See the evidence

Agentic systems and tool calling

Agents that answer only through verified tools: schema fetched before any filter is built, arithmetic computed in code and narrated by the model, multi-agent orchestration over MCP.

See the evidence

Real-time voice and multimodal

Full streaming chains: speech to text, reasoning with tools, text to speech, at conversational latency, on local models or APIs. Plus the unglamorous layer that makes it a product: SIP trunking and telephony.

See the evidence

Executive analytics and information architecture

Operating tools that close a loop: detect, diagnose, recommend, assign, measure. Not dashboards that decorate one.

See the evidence

Enterprise and constrained deployment

Private endpoints, VNet integration, data residency, internal-only model gateways, offline-first requirements. Constraints treated as design inputs, not obstacles.

See the evidence

The first step

Twenty minutes. You describe where AI is stuck.

You leave the call knowing whether I can help, roughly what it would take, and what it would cost to find out for sure. If I am not the right person, I will say so on the call.

The 20 minutes, in order

  1. You talk first. Where AI is stuck, what has been tried, what it costs today.
  2. I answer plainly. Whether I can help, and what I would look at first.
  3. You leave with a next step. An audit scope, a pointer elsewhere, or a clean no.

Before you book

  1. Your stack is not too messy to start. Messy is the normal starting condition.
  2. Training that does not survive the week is the normal outcome. These sessions build on your backlog and ship something real.
  3. You do not need budget approved to take the call. You need it approved to start step two.
Book the 20 minutes Not ready. Send me the newsletter