Learn the work behind Data, AI & Forward Deployed Engineering
Practical explanations, career decisions and reproducible workflows. Read the reasoning, inspect the evidence and follow the next skill into a real programme.
Build a per-client AI cost attribution model
A provider invoice cannot explain which client workflow creates value or waste. Cost attribution needs stable client, run and task identifiers across model calls, tools, storage and retries.
Build a point-in-time correct feature join
A standard join can attach the latest feature ever recorded, including values created after the prediction. A point-in-time join selects the latest snapshot for the same entity whose feature timestamp is at or before tha
Build a portfolio around evidence rather than screenshots
A screenshot proves that one screen existed. A reviewer needs to see the problem, decisions, executable checks and recovery path.
Build a portfolio evidence index for interviewers
A reviewer should not need to search an entire repository for the split, metric or reproduction command behind a claim. An evidence index turns the project landing page into a map from statement to artifact.
Build a product funnel with an explicit event contract
Define entry selection, step order, conversion window, counting unit and observation maturity before calculating a funnel. Counting users who ever fired two event names does not establish that they completed a valid sequ
Build a professional FDE portfolio from anonymized evidence
Professional experience is valuable evidence, but a public portfolio must not expose client identity, records, credentials or confidential measurements. Strong anonymization often means recreating the proof rather than r
Build a proper date table for business reporting
A date table represents the reporting calendar independently of transactions. It provides one row per date and attributes such as year, month and fiscal period, so reports can group activity consistently and represent da
Build a Python CLI that another person can install
A script that works from its author’s folder is not yet a distributable command. An installable CLI declares metadata, Python compatibility, dependencies and an entry point.
Build a queue-backed task with visible status
Long work should not leave an HTTP connection open indefinitely. Accept a task, return its identity and expose a state machine the client can poll or receive by callback.
Build a RAG document inventory with source ownership
A retrieval system cannot manage what it has never inventoried. Before chunking or embedding, record where each source came from, who can approve it, which version applies and who may retrieve it.
Build a read-only business-metrics tool contract
A read-only metric tool is a safer first MCP exercise than a generic database query. It can expose approved definitions and aggregates while keeping SQL, credentials and row-level data behind the service.
Build a reconciliation step into every automated report
An automated report should explain how source records become selected records and how selected values become displayed totals. Reconciliation turns that explanation into executable checks that can stop a release when the
Not sure which programme fits?
Tell us your background and we will map it to the right entry point — including saying so when a cheaper programme is the better fit. A counsellor replies within one working day.