Senior AI Data Scientist – Agentic Process Development
Posted 1ds ago
Employment Information
Report this job
Job expired or something wrong with this job?
Job Description
Senior AI Data Scientist building trustworthy agentic systems to automate team.blue’s multi-country HR processes. Developing production integrations, safeguards, evaluations, and cost-efficient LLM workflows.
Responsibilities:
- Map HR processes end to end across multiple countries and quantify FTE cost per step
- Facilitate workshops with Talent, Rewards, and HR Ops leaders to score automation candidates
- Extract requirements from senior non-technical stakeholders
- Design state transitions, triggers, branches, system calls, waits, escalations, and failure handling
- Build guardrails including dry-run mode, approval gates, least-privilege credentials, and rollback paths
- Decide where humans remain in the loop and establish confidence thresholds and usable review queues
- Create evaluations for task completion, hallucination, and gendered or culturally biased language
- Build webhook-based, idempotent agent integrations
- Optimize model selection and routing based on cost and latency
- Evaluate HR-tech vendors and their APIs, data models, extensibility, and integration costs
- Learn HR-tech systems, HR-data governance, employment-law constraints, EU AI Act obligations, and organizational change management
Requirements:
- 7+ years building data and ML systems in industry
- Experience spanning both sides of the LLM shift
- Experience shipping systems with permission to take irreversible actions affecting real customers
- Expert in Python and ML
- End-to-end production ownership, including readable Python, containerisation, and instrumentation
- Current toolchain experience such as uv, Docker, or equivalent
- Production experience with multi-step, tool-calling LLM workflows, including orchestration, retries, idempotency, timeouts, and partial-failure recovery
- State-machine design experience
- Cost and latency engineering, including model routing, caching, and batching
- Safety practices including staging modes, approval gates, least-privilege scoping, and rollback
- Evaluation design for generative and agentic output, including LLM-as-judge, golden-transcript regression suites, and red-teaming
- Applied statistics
- Process mapping and quantification
- Ability to facilitate workshops with senior non-technical stakeholders
- Executive-grade written business cases and cost modelling for Finance audiences
- Technical vendor evaluation covering API surface, data model, extensibility, and integration cost
- Proof of eligibility to work in the country applied for
- Nice to have: Master's or PhD in Computer Science, AI, Machine Learning, or a related field
- Nice to have: Existing EU AI Act / high-risk-AI-system familiarity
- Nice to have: PromptOps at scale
- Nice to have: DataOps/MLOps practice
- Nice to have: Prior exposure to HRIS/HR-tech or automation in an audited or regulated domain
- Relocation packages and visa sponsorship are not supported
Benefits:
- Diversity & Inclusion commitment
- Respectful, open, and trusted collaboration culture
- ESG and sustainability initiatives
















