Functional Analyst

Liquidity Risk Developer (Python / SQL) | Freelance | Brussels

The project

A Belgian bank's ALM & Liquidity Risk team is rebuilding its two main liquidity stress-testing tools.

The team sits in the second line of defence, within Risk. Both tools were built in-house over many years in Excel, Access and VBA, and they now carry a large part of the team's regulatory work. They have reached their limits on speed, efficiency and governance.

The bank has the budget to build a version 2.0, and needs one hands-on consultant to lead it.

The two tools

Dynamic balance sheet tool: runs macro scenarios (yield curve, inflation, GDP, new business volumes) to produce a stressed balance sheet over 5 years, then calculates liquidity metrics such as LCR.

Snapshot tool: takes the balance sheet at one point in time and calculates a 3-month forward liquidity metric, similar to LCR but with a 3-month horizon instead of one.

Both are linked to the bank's ILAAP.

Phase 1: Assess and design (about 2 months)

Analyse both tools: their logic, data flows, users and dependencies.

Document the use cases and regulatory requirements, with the ILAAP in mind.

Decide on a target design, including whether it should be one tool or two.

Recommend the technical approach. The expected stack is Python and SQL.

Phase 2: Build (about 2 months to start)

Take the first steps of the rebuild, hands-on.

Put in proper governance: controls, access rights, execution logs and documentation.

Set it up so the team can take it over and keep developing it.

What you'll need

Hands-on experience writing production-quality Python and SQL yourself, without relying on AI coding assistants.

Experience in liquidity risk or ALM: stress testing, LCR, NSFR, ILAAP or balance sheet projection.

Experience reading, picking apart and rebuilding Excel, VBA or Access tools.

A track record of putting governance and controls into data or risk processes. BCBS239 exposure is a plus.

Able to work on your own and explain technical choices to a business audience.