Benefits Rules-Engine Modernisation Director

The mandate

Benefit eligibility and payment calculations rely on decades of code, parameter tables, manual instructions and appeal precedents that no single team can reconcile. Policy changes take months because engineers must infer intent from implementation, and similar households sometimes receive different outcomes depending on which exception path executes. The defined problem is to design a modernisation route that improves explainability and speed without silently rewriting lawful entitlements during technical decomposition.

The deliverables are a rule-source concordance, decision-domain map, exception and precedent model, target rules architecture, migration test corpus, sequencing plan and investment dossier. The design must separate legislation, formal policy, discretionary guidance, data derivation and operational judgement; preserve effective dates and household context; and show how a citizen, caseworker, appeal body and auditor can reconstruct why a result occurred before and after migration.

Four milestones span six months: week five accepts the provenance and decision-risk baseline; week twelve approves bounded rule domains and the target explanation contract; week nineteen completes comparison tests across representative and adversarial citizen cases; and week twenty-six delivers migration waves, control design, capability plan and funding choices. Billing attaches to acceptance of each milestone’s specified evidence.

The director-general will accept the work when policy owners can trace sampled rules to authoritative sources, caseworkers can understand proposed explanations, legal advisers confirm the interpretation process, and legacy versus target results are classified rather than averaged away. The dossier must identify protected cohorts, appeal impacts, backdating, manual judgement and rollback, and it must enable procurement or internal build decisions without consultant-dependent architecture translation.

The client provides legislation and policy repositories, parameter histories, source-code access, anonymised case and appeal files, operational manuals, change requests, defect history, data lineage and bilingual subject experts. The consultant does not make eligibility decisions, reinterpret law, alter live rules, select a technology supplier, procure delivery, process citizen data outside the secure environment or run the subsequent implementation.

Why this is external work

Policy, legal, operations and technology each hold a different fragment of why the current engine behaves as it does, and prior programmes began from preferred products rather than contested decision meaning. The agency also needs to challenge discrepancies without assigning blame for historic implementation. Independent expertise can create a neutral rule lineage, force explicit treatment of judgement and appeal, and give investment leaders a route grounded in citizen consequences.

What you will own

• Reconcile legislation, policy, guidance, parameters, code, manual practice and appeal precedent for the highest-risk decision domains.
• Classify exceptions by lawful basis, effective period, affected cohort, operational discretion, frequency and consequence of misapplication.
• Design bounded rule services with versioning, provenance, explanation, human-judgement and cross-domain dependency made explicit.
• Construct a migration corpus spanning ordinary, edge, historic, backdated, multilingual and previously appealed citizen cases.
• Define comparison outcomes that distinguish corrected legacy defects, intended policy change, data difference and unexplained divergence.
• Sequence domains using citizen harm, change demand, interpretive confidence, coupling, operational readiness and rollback feasibility.
• Deliver the target blueprint, evidence standard, delivery options, capability needs and decisions reserved for public officers.

Candidate qualifications

• Has modernised a complex statutory rules engine in benefits, tax, insurance, pensions or another rights-bearing decision environment.
• Can separate legislative authority, policy intent, parameterisation, code behaviour and caseworker judgement without erasing ambiguity.
• Has built migration test corpora that reveal cohort harm and historic exceptions beyond ordinary regression-test coverage.
• Understands explainable decision records, effective dating, appeals, backdating, bilingual policy interpretation and protected citizen information.
• Has guided build, buy and incremental decomposition choices without allowing a preferred rules product to dictate the domain model.
• Produces evidence usable by policy officials, lawyers, case operations, technologists, auditors and public investment governance.

Non-negotiables

• Can work the Brussels hybrid cadence and conduct approved caseworker observations and bilingual design sessions.
• Must disclose relationships with rules-engine vendors, systems integrators, public agencies and prospective procurement participants.
• Brings rights-bearing decision modernisation; generic workflow automation or low-code implementation does not satisfy the mandate.
• Will keep citizen data in the secure client environment and leave every policy interpretation to authorised public officials.

1. 49 words maximum. Describe a legacy rule discrepancy that was lawful policy ambiguity rather than defective code.
2. 49 words maximum. Which adversarial citizen case would you place in the migration corpus first, and why?
3. 49 words maximum. How would you make a decomposed eligibility decision reconstructable during an appeal?