What I actually do.
Three tracks, in the order I would claim them. AI consulting is the work; engineering is how it ships; go-to-market is what I can speak to rather than sell.
Making AI useful, not just present.
Underneath all of this is ordinary model-level fluency — how these systems fail, what they cost, and when a smaller one is the right answer.
LLM Systems
Retrieval, prompting and tool use, built into a working system.
RAG
Chunking, embeddings and reranking tuned to your own corpus.
AI Agents
Agents that take real actions, with the guardrails to be trusted.
Evaluation
Test sets and metrics that show whether a system is improving.
AI Governance
Policy, access and audit trails for systems that touch real data.
Opportunity Mapping
Finding where AI genuinely helps, and where it plainly does not.
- Retrieval
- Embeddings
- Reranking
- Tool Use
- Guardrails
- Evaluation
- Fine-tuning
- Prompt Design
- Context Windows
- Observability
- Cost Control
- Latency
The stack behind meaningful work.
Web and mobile applications, the services behind them, and the infrastructure that keeps them running — built to be handed over, not just delivered.
Web Apps
Production front ends in React and Astro, typed end to end.
Mobile Apps
React Native applications that ship to both of the app stores.
APIs & Data
REST and GraphQL services, and the data models behind them.
Infrastructure
Containers, CI and cloud deploys that survive a bad Friday.
Testing
Unit, integration and end-to-end suites people actually run.
Security
Auth, secrets and dependency hygiene handled before launch.
- TypeScript
- React
- React Native
- Node.js
- Python
- Postgres
- Astro
- Tailwind
- Docker
- GitHub Actions
- Playwright
- OpenTelemetry
Also fluent in the commercial side.
Not a practice I sell — the part of the job that decides whether good work ever reaches anyone, and which I can hold a real conversation about.
Positioning
Saying what a product is, for whom, and against what else.
Demand Gen
Channels and campaigns measured on pipeline, not impressions.
Sales Process
Stages, criteria and handoffs that a team can actually follow.
Discovery
Question sets that surface the problem behind the request.
Enablement
Materials that let someone else run the conversation well.
Revenue Analytics
Funnel and cohort reporting that survives a hard question.
- Positioning
- ICP Definition
- Messaging
- Pipeline
- Forecasting
- Onboarding
- Retention
- Pricing
- Segmentation
- Attribution
- Cohorts
- Churn
