Head of AI
HybridSingapore, Singapore
Job Summary
Define TSC's AI technical strategy across stakeholder mapping, classification, summarization, entity resolution, risk detection, and workflow automation. Own the end-to-end lifecycle of material AI capabilities, from data foundations and evaluation design through deployment, monitoring, and incident response. Build evaluation systems including golden datasets, regression tests, and human-review thresholds while establishing governance for model agents, prompt injection, and data leakage. Architect production systems using LLMs, retrieval, and deterministic orchestration to balance customer value, quality, speed, and unit economics. Lead a team of AI and Data Engineers, partner with Product and Platform Engineering, and explain architecture trade-offs to executives. Establish repeatable agentic workflows with explicit permissions, stopping conditions, and cost limits to improve delivery velocity without uncontrolled autonomy.
Required Qualifications
- Typically 8+ years building production software, data or AI systems
- Exceptional evidence of production delivery
- A proven record of shipping and operating production AI or data systems
- Practical depth in LLM systems
- retrieval-augmented generation
- orchestration
- embeddings
- classification
- summarisation
- extraction
- evaluation
- Strong experience with data pipelines
- Experience with cloud data platforms
- Experience designing systems for enterprise or sensitive-data environments
- The ability to review code, data models, system architecture and evaluation results in depth
- The ability to explain AI strategy, architecture, quality, cost, risk and roadmap trade-offs to executives, customers and non-technical stakeholders
Desired Qualifications
- Experience with GCP
- BigQuery
- AlloyDB
- Vertex AI
- comparable platforms
- designed evaluation, quality-control or regression systems for AI products
- built or governed agentic workflows with explicit permissions, stopping conditions, validation gates and human escalation
- managed model quality, cost and latency trade-offs in production
- worked closely with Product to turn ambiguous customer workflows into measurable, shippable AI features
- improved retrieval quality, entity resolution, classification or structured extraction over messy real-world datasets
- led teams from manual prompting and experimentation toward repeatable, observable and governed AI workflows
- supported enterprise customer conversations involving AI reliability, security, architecture or limitations
Hiring someone like this?
Get your role in front of qualified candidates on Sorce.