AI / ML Engineer at K3 Advisory Group

Kuala Lumpur, Kuala Lumpur, Malaysia -

Full Time

Start Date

Immediate

Expiry Date

22 Aug, 26

Salary

165000.0

Posted On

24 May, 26

Experience

2 year(s) or above

Remote Job

Yes

Telecommute

Yes

Sponsor Visa

Skills

Python, Pandas, LLM Application Development, Retrieval-Augmented Generation (RAG), SQL, MLOps, Prompt Engineering, Tool Calling, Model Monitoring, Vector Databases, Azure AI Services, Semantic Search, Data Security, NLP, Financial Analysis, Responsible AI

Industry

Business Consulting and Services

Description

About K3 Advisory Group K3 Advisory Group is a UK professional services group with 18 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 shared platforms, data foundations and AI-assisted products that scale across these businesses while accommodating local variation. Engineers in this team work close to commercial outcomes, with direct visibility of advisors, partners and clients. Role Purpose Design, build and operate practical AI capabilities that help K3 users interrogate business data, automate analysis and reduce manual effort — while preserving data security, auditability and human oversight. The role is explicitly about applied AI in a regulated professional services context, not research. Success looks like AI-assisted features that advisors use day-to-day, that ground their answers in approved data, that fail safely, and that can be defended in front of internal audit, compliance and (where relevant) regulators. Key Responsibilities AI Capability Engineering • Build AI-assisted features using approved model providers, orchestration frameworks and internal platform patterns. • Design safe tool-calling workflows over governed data sources, including read-only query tools, metric lookup tools, and schema/context tools. • Support document intelligence, financial analysis, variance explanation, forecasting support and guided Q&A use cases. • Develop model and prompt versioning, telemetry, model monitoring and usage reporting. Evaluation, Guardrails & Safety • Implement prompt, tool and model evaluation suites to test answer quality, safety, tenancy and regression risk. • Build guardrails that prevent models from choosing unauthorised client context, accessing raw data outside approved contracts, or exposing sensitive values in outputs. • Define and maintain golden datasets and test scenarios that reflect real K3 workflows across different subsidiaries. • Operate kill-switches, rate limits, fallback paths and human-in-the-loop checkpoints for higher-risk use cases. Data & Platform Collaboration • Work with data engineers to define semantic contracts, metric definitions and approved analytical functions that AI features can rely on. • Work with full-stack engineers to integrate AI features into web applications, dashboards and workflow screens. • Contribute to the Group's responsible AI standards, including documentation of model cards, intended use, known limitations and review history. Stakeholder Engagement • Explain model behaviour, limitations and risk controls to non-technical stakeholders — including partners, compliance and risk functions. • Translate business questions into measurable AI capabilities with defined inputs, outputs and evaluation criteria. • Provide pragmatic guidance on where AI is and is not the right tool, including when conventional analytics or process automation would be safer or more effective. Required Experience & Skills • 3–5+ years applying AI, ML or data science techniques in production or business-critical environments. • Strong Python experience, including pandas and modern ML/AI libraries. • Experience with LLM application development, including tool calling, retrieval-augmented generation (RAG), evals, prompt management or agent-style workflows. • Experience with SQL, analytical datasets and governed data access patterns. • Familiarity with model monitoring, experiment tracking, model registry or broader MLOps practices. • Understanding of privacy, security, audit logging and data minimisation in AI systems. • Ability to translate business questions into measurable AI capabilities and clearly explain limitations to non-specialists. Desirable Experience • Experience with document intelligence, NLP, financial data, forecasting or classification problems. • Exposure to regulated environments (financial services, legal, audit) and an appreciation of the additional controls this demand. • Experience with vector databases, semantic search, or hybrid retrieval architectures. • Familiarity with OpenRouter, Azure AI services, Azure OpenAI, or comparable enterprise AI tooling. • Experience contributing to responsible AI policies, AI risk assessments, or model governance forums. Success Measures • User impact: AI features measurably improve user productivity without weakening access control or auditability. • Grounding: AI answers are grounded in approved data sources and tested against known scenarios — hallucinations on material questions are caught before release, not after. • Regression control: Evals reliably catch regressions in tenancy, permissions, financial accuracy and response quality before features are promoted. • Transparency: Model and tool usage, failures and limitations are visible through telemetry and reviewed regularly by engineering and compliance. • Defensibility: AI features can be explained, evidenced and defended to internal audit and, where relevant, external regulators. Working Environment • Reporting line: Group Technology leadership, working day-to-day within a cross-functional product squad (engineering, data, AI, design, product). • Stakeholders: UK-based business leaders, partners, operational teams, compliance and risk functions, and client-facing advisors across multiple subsidiaries. • Delivery model: Iterative, product-led delivery with short feedback loops, paired with the governance discipline appropriate to a regulated professional services environment. • Tooling baseline: Modern cloud platform (Azure-first), Git-based source control, CI/CD pipelines, infrastructure-as-code, observability tooling and a documented engineering handbook. • Ways of working: Code review, pairing, design reviews, threat modelling for sensitive features, and lightweight architecture decision records (ADRs). Governance, Security & Compliance Expectations Every engineer in Group Technology is expected to treat the following as non-negotiable foundations, not optional extras: • Confidentiality: Client, matter, and case data is highly sensitive. Need-to-know access is the default; broad access is the exception and must be justified. • Security by design: Threat modelling, secure defaults, secrets management, dependency scanning and least-privilege access are built into features from day one. • Responsible AI: Where AI is used, model behaviour, prompts, tools and data access are versioned, evaluated and monitored. Human oversight is preserved for material decisions. • Regulatory awareness: For features touching FCA-regulated entities (e.g. Pareto, Luna), additional controls apply around record-keeping, client communications and data handling. Engineers are expected to flag uncertainty early. • Data protection: UK GDPR and Group data protection standards apply across all subsidiaries; data minimisation, lawful basis and retention controls are part of normal design. Development & Progression • Clear engineering career path with senior, lead and principal levels, plus a parallel route into architecture or engineering management. • Exposure to acquisitions, integrations and greenfield product builds across multiple professional services disciplines. • Supported learning budget, certifications relevant to the role, and time allocated for proof-of-concept work and tooling improvements. • Direct line of sight to commercial outcomes — engineers see how their work changes how advisors and clients actually operate. Person Specification • Sceptical: Comfortable challenging proposed AI use cases when a simpler, safer or more auditable approach would serve users better. • Communicative: Explains model behaviour and limitations in language that partners and compliance teams can act on. • Collaborative: Works closely with data engineering, full-stack engineering, product and risk functions to deliver AI features that stick.

Responsibilities

Design and operate practical AI capabilities to automate business data analysis and reduce manual effort within a regulated professional services context. Build safe tool-calling workflows and implement evaluation suites to ensure answer quality, security, and auditability.