Jul 2025 — present / Internal pilot · toward production
LexisNexis — Applied AI
Making fraud-model workflows recoverable and reviewable.
During my placement I built a working demo for fraud-model optimisation. I was retained as an AI engineering contractor to develop the capital-funded internal pilot toward production.
01The context
Fraud-model optimisation combines analysis, modelling and decisions that need accountable human review.
02My contribution
Initiated a working agentic prototype during my placement, then stayed on as a contractor to develop the funded internal pilot.
03The approach
- Custom Python orchestration and reusable Copilot skills coordinate work with persistent SQLite state.
- Audit trails and mandatory review gates keep human judgement in the workflow.
- DVC and Papermill make notebook-based work recoverable; LightGBM discovers candidate rules, with constrained logistic regression for policy weighting.
04Evidence & limitations
- Working demo delivered to the AI engineering team by the end of the placement.
- Retained as an AI engineering contractor in August 2026 to develop a capital-funded internal pilot.
- Two fraud-model optimisations completed during the earlier placement.
The boundaries of this account
Employer-confidential work: no source code, transaction data or internal evaluation results are published.
This describes a pilot building toward production, not a launched production platform.
Descriptions are owner-supplied; no independently verified performance uplift is claimed.
05Tools & methods
- Python
- SQLite
- DVC
- Papermill
- LightGBM
- Snowflake
- PyTorch
Continue exploring
TypeForge