About the company:
Our client is a global, science-driven biopharmaceutical company (top 10 worldwide)
focused on discovering, developing, and delivering innovative medicines and
healthcare solutions that improve patients’ lives. They are heavily investing in AI and
are building advanced Agentic AI solutions to accelerate scientific research and
decision-making across the organization.
Brief Job Profile Description
The role holder designs, builds, and owns AI agent workflows that run in real business processes for Oncology Biometrics, working within the AI Strategy and Innovation (OBM AI) team. Takes workflows from prototype to monitored, evaluated production services with a usable front end, and is accountable for their reliability, cost, and performance after release. Will mentor junior engineers and work with a cross-functional network of technical experts to deliver reliable, compliant AI solutions.
Typical Accountabilities
- Design agent and multi-agent workflows end to end, including task decomposition, state and memory, tool design, routing, failure recovery, and human-in-the-loop checkpoints.
- Build production backend services (Python, FastAPI), RAG pipelines, and MCP/API integrations with enterprise systems, and deliver the user-facing application, directly or with front-end engineers, including human review-and-approve steps.
- Build evaluation and observability for each workflow (evaluation sets, quality metrics, regression testing, tracing, cost and latency monitoring) and apply guardrails against hallucination, prompt injection, and tool failure.
- Deploy and operate services in production (containers, CI/CD, cloud), optimize model choice and routing for cost and latency, and follow up on incidents.
- Deliver to GxP-aware standards, including traceability, audit trails, access control, versioned prompts and models, validation-supporting documentation, and human review, in line with AstraZeneca standards, responsible AI, and data privacy requirements.
- Review code and designs, mentor junior engineers, and contribute reusable components and patterns to the wider team.
Qualifications and Skills
- Essential: Degree in computer science, engineering, or a related discipline; 4–7+ years of software engineering, including 1–3 years building LLM or agentic systems; has shipped at least one agentic workflow into production use in a real business process (real users, operated after launch) and can describe what broke; deep, hands-on Python; production experience with LangGraph, LangChain, or an equivalent framework, RAG, and LLM APIs; demonstrated evaluation and observability practice; experience deploying services (containers, CI/CD, cloud); practical understanding of GxP-aware delivery.
- Hands-on agentic AI experience. Someone who has built application with agent workflows, using tools such as Claude Core or Codex, and can help design and build our internal AI systems
- Leadership: Someone who has led application development or small engineering team, working closely with business stakeholders.
- Desirable: Fine-tuning (LoRA/QLoRA), model routing, or multimodal agents; Kubernetes and a major cloud AI service (Azure AI Foundry, AWS Bedrock, or GCP Vertex AI); React/Next.js and TypeScript; knowledge of 21 CFR Part 11, EU Annex 11, and computer system validation; clinical or statistical data experience (SAS/R/Python, CDISC); open-source contributions to agent or evaluation tooling.
Key Relationship to Reach Solutions
- Internal: OBM AI Strategy and Innovation leadership and engineering team; Biometrics process owners and users of the workflows; R&D IT; Enterprise AI; Oncology Data Science & AI; Quality and Compliance Teams
- External: AI and technology vendors and external service providers; open-source and developer communities
