K3 Advisory Group is a UK professional services group with multiple trading subsidiaries and more than 1,200 staff. The Group spans corporate finance, tax, restructuring and insolvency, legal, financial planning, and technology-enabled advisory services. All technology delivery must balance the pace required to exploit AI, data and automation with strict expectations around governance, regulatory obligations, security-by-design, audit trails, and client confidentiality.

Group Technology builds the shared platforms, data foundations and AI-assisted products that scale across these businesses while accommodating local variation.

Our AI Intelligence Layer

K3's AI Intelligence Layer sits beneath our Salesforce CRM and Salesforce Marketing Cloud, turning company and deal data into scored, explainable signals our advisors and Growth Operations team can act on.

Role Purpose

Own the build and governance of this machine layer end to end — the schema and data foundations beneath it, the statistical / ML models it runs, and the middleware that connects it to Salesforce. You'll be the named model owner for every statistically / ML-driven use case in the programme, accountable for accuracy, explainability and governance sign-off before any model is relied on commercially.

Key Responsibilities

Schema & data foundations

• Own and extend the identity / firmographic and financial layers of the ICP Profile Schema, working from Companies House API data and the Data Platform.

• Build and maintain rules-based enrichment and de-duplication.

Statistical & ML model build

• Build, validate and maintain supervised classifiers and regression models: succession risk score, distress indicator model, exit-readiness score, win-probability model, and deal-value estimation.

• Own model recalibration on a scheduled cadence, maintaining documented accuracy and precision thresholds for each model before any commercial reliance is permitted.

Middleware ownership

• Own the middleware layer connecting the ICP Profile Schema to Salesforce (CRM) and Salesforce Marketing Cloud (campaign engine) — the layer every other layer reads from and writes to.

• Ensure every model output carries a confidence score and reason code, to the explainability standard defined for our AI output layer.

Governance

• Act as named model owner across the Group's three-tier AI / model governance framework — accountable for validation, change control and explainability at each governance checkpoint.

• Design and build to the standard required by our FCA-regulated entities from day one, even though this phase sits entirely in non-regulated businesses.

• Take all middleware and model changes through the Architecture Board gate.

• Report model and use-case status through the Group's existing technology governance reporting for the AI Intelligence Layer workstream — this isn't run as a separate governance track.

Required Experience & Skills

• Proven delivery of supervised ML models (logistic regression, gradient boosting) and deterministic rules engines in production, run side by side.

• A track record of building explainable model outputs — reason codes, confidence scoring — for governed or regulated environments.

• Comfortable owning a named model-risk position: able to defend model accuracy and logic to a governance committee, not just to engineering peers.

Desirable Experience

• Exposure to regulated-data handling (financial services, FCA or equivalent), given the design-for-regulation requirement.

• Experience integrating ML or rules-engine output with Salesforce or a comparable CRM / marketing platform.

• Familiarity with Companies House or equivalent company-data APIs, and firmographic / financial data enrichment.

Success Measures

• Accuracy & explainability: every live model meets its documented accuracy / precision threshold and carries a confidence score and reason code before commercial use.

• Governance: every use case has an up-to-date, signed-off governance record; no model or middleware change bypasses the Architecture Board gate.

• Delivery: schema, model and middleware changes ship against the AI Intelligence Layer roadmap, with clean status reporting into the Group's technology governance pack.

• Scalability: the platform is proven fit to extend from the proof of concept to further verticals without rework.

Governance, Security & Compliance Expectations

• Confidentiality: company, deal and firmographic data is commercially sensitive; need-to-know access is the default.

• Explainability by design: every model output is versioned, evaluated and carries a confidence score and reason code before use.

• Change control: all middleware and model changes go through the Architecture Board gate and the Group's three-tier AI / model governance framework.

• Data protection: UK GDPR, Malaysian PDPA (where applicable) and Group data protection standards apply.