$graph ./stack
An ecosystem, not a logo wall
These technologies are connected by the products I shipped with them. Trace any node to see what it touches.
The stack, top to bottom
How a request travels through what I build
The same tools, as a network
What connects to what
Shape of the toolkit
Deep where it counts, honest about the edges
Retrieval and evaluation are where I go deepest. Infrastructure is the axis I am still filling in, and I would rather say so than pad the list.
retrieval
evaluation
infra
Product and ownership
The part of the job that happens before any code exists, and after it ships.
- Problem framing and scopingDaily
- User interviews and discoveryStrong
- Roadmap and prioritisationStrong
- Metric definition and instrumentationDaily
- Pricing and packagingWorking
- Writing specs and decision recordsDaily
AI and ML
Retrieval, agents, and the evaluation work that makes either of them trustworthy.
- Retrieval systems (BM25, dense, hybrid, RRF)Daily
- Evaluation (nDCG@k, MRR, Recall@k, LLM-as-judge)Daily
- Cross-encoder rerankingStrong
- LLM agents and orchestration (LangGraph)Strong
- PyTorch and TensorFlowStrong
- Weights & Biases experiment trackingStrong
Full stack engineering
Everything between the database and the pixel, shipped by one person when it needs to be.
- TypeScript and ReactDaily
- Next.js (App Router)Daily
- Python and FastAPIDaily
- PostgreSQL and pgvectorStrong
- Node.js APIsStrong
- Interface design and design systemsStrong
Cloud, data and tooling
Deploying it, watching it, and knowing when it breaks.
- AWS (Lambda, S3, EC2, SageMaker, CloudWatch Logs, Route 53)Strong
- GitHub Actions and CI pipelinesDaily
- DockerStrong
- Observability (OpenTelemetry, Sentry)Strong
- Vercel and SupabaseStrong
- Stripe and payments integrationWorking
$./next-step
Need one of these on your team next quarter?
I am most useful where product thinking and AI systems overlap, and where somebody has to own both.