Projects
Real things to build, each with a page that says why it matters, what it looks like, what it's built with on each cloud, and exactly what needs doing next. Some are under way and you can join; some haven't started and need a lead.
To join: open a project and use Join this project, or pick a good first task from its repository.
More projects live in their own sections: AI for a cause and Deep tech.
Conversation UI on Data
A specification for chat-with-your-data where the model picks certified metrics and never writes SQL.
Chat and dashboards read the same semantic layer, so they always give the same answer.
DataEngineering Training: Ecom Simple Data Platform
Learn data engineering by designing real projects; project 1 is a small e-commerce lakehouse on Databricks.
Learners practise the write patterns (append, overwrite, merge, history) that real pipelines use daily.
LLM Cost Explainer
Explains why LLM spend changed and which mistakes cost money, from a provider usage export alone.
Rules point to fixable mistakes and estimate what each fix would save.
LearnLLMStack
Build a tiny version of every LLM layer, from your own model up to the agent, and connect them.
Engineers who know why reading a prompt is cheaper than writing an answer make better choices about context size and caching.
LearnLangGraph
A hands-on LangGraph course in three waves: graphs without an LLM, then agents, then production patterns.
A new engineer can go from zero to a tool-using agent with memory in a few evenings.
Ontology and Knowledge Graph
Build an ontology and knowledge graph from scratch over code, data, specs and AI assets.
Impact analysis shows every downstream table, screen and team before a change ships.
Spark Cost Agent
A Spark cost analyst agent built as a lab for agent tracing, evaluation and cost per successful task.
Cost per successful task shows whether an agent is worth its spend, not just total tokens.
AI Assurance Catalog
An open list of what must be true of an AI application before it is trusted in production.
Test results become a coverage report that maps to ISO/IEC 42001, NIST AI RMF, OWASP and the EU AI Act.
AI Harness Catalog
An open list of the parts that must exist around a model call for production-grade AI.
Every capability states the failure that follows when it is absent, so gaps are found at design time.
AgentTwin
Test an AI agent in a simulated world and judge what changed, not only what it said.
Rule-breaking writes, such as a refund over a limit, are found in simulation, not in the ledger.
AgentTwin Lab
Simulated days for AI agents, to see whether automated fixers weaken the checks that caught them.
Each change a fixer makes or tries is labelled as weakening or not, before it can ship.
Claims FNOL assistant on Azure
A motor-insurance claims agent on Azure, built on the same harness as the support agent.
Claims are registered at once and payouts up to the limit are paid without waiting for a handler.
Clinical-trial data review agent
An agent that finds impossible or inconsistent patient data in a clinical trial and drafts the query sent to the site.
The agent reads the whole record and proposes only the queries worth sending.
Eval Harness Framework
One framework to check AI agents with any provider's evals, guardrails, cost and business rules.
One adapter per provider replaces the custom code teams write each time they combine two eval tools.
Reference Agent
An open customer-support agent built and tested to the catalogs, layer by visible layer.
Teams start from a support agent that already passes behaviour, feature, structure and harness gates.
The Agent Harness course
A hand-drawn course following one real agent from a customer's message to the real code.
New team members learn a production agent's design in hours instead of reading code for weeks.
AskSherlock
An open-source Genie-style chat over a ClickHouse cost store, where the model never writes SQL.
Chat and dashboards share one set of metric definitions, so they cannot disagree.
MyClaudeCode
A mini coding agent used as an instrument to put numbers on unmeasured coding-agent advice.
Knowing which always-loaded layer dominates the baseline lets a team cut it first.
AI job-loss pain registry and helpers
Evidence-backed problems of people losing work to AI, and simple free tools that help them earn again.
People avoid scams and claim entitlements they would otherwise miss.
Drug-safety signal investigator on Databricks
A pharmacovigilance agent built only on Databricks' own AI services, from public FDA data.
Report counts and disproportionality statistics are computed in code, never by the model.
Clean Data Engineering specs
Two open catalogs for data platforms: what parts must exist, and what must be true of the data.
Teams can see in one place what their platform is missing before it bites them.