AI Technical Lead; Microsoft + Python), AI-Augmented Pods

Central Europe

MICROSOFT (hard gate)

• 5 + years on the Microsoft stack, with real depth in at least two of:
  Dynamics 365 and Dataverse, Power Platform (Power Automate, Power Apps, Power Pages, environments, DLP, ALM), Azure application development.
• Azure delivered inside a customer tenant under their governance: managed identities, Key Vault, Functions or Container Apps, VNets and private endpoints, and a security review passed rather than worked around.
• Azure Dev Ops end to end, including standing up CI/CD from nothing.
• Copilot Studio or Azure AI Foundry in a real client deployment, not a sandbox.

• Production Python written in the last two years, and the ability to review someone else's Python. Typing, tests, packaging, async. Scripting does not count.
• At least one LLM or agent system taken to production and kept running: tool and function calling, MCP, orchestration and state, fallbacks, human in the loop.
• Evaluation discipline. Has built golden sets from real cases, run regression suites, measured against a baseline, and used the numbers to change or stop a workflow.
• Observability for LLM systems: traces, token and cost telemetry, alerting.

• Has led engineers technically, not only managed them: set standards, reviewed code, mentored, owned architecture decisions and lived with the consequences.
• Can hold an architecture conversation with a client's security and architecture function and come out with an approval.
• Says no with a reason and a written alternative. Keeps a decision log.

• Be able to travel to Rotterdam and other client sites in two-day blocks.
• English C1, spoken and written.
• Available at half allocation or hourly, and comfortable holding two or three accounts at once.

• Financial services or another regulated domain: asset management, insurance, payments, pharma or life sciences.
• Snowflake specifics:
  Cortex, semantic models, dbt, MCP over Snowflake.
• Microsoft certifications: PL-600, PL-400, AI-102, AZ-204, AZ-305, DP-203.
• Semantic Kernel, Auto Gen or an equivalent orchestration framework.
• Power Platform governance at scale: environment strategy, DLP, managed solutions, centre of excellence toolkit.
• Experience turning business-user scripts and citizen-developer tooling into governed services.
• Compliance evidence work: DORA, EU AI Act, ISO 27001, SOC 2.
• Knowledge graph and ontology work, Neo4j, data lineage tooling.
• C# or .NET for integration work inside Microsoft estates.
• Presales support: shaping the technical answer in an RFP, running a discovery session, defending an architecture in front of a buying committee.
• Written enablement: playbooks, runbooks, training plans, certification paths.

ARCHITECTURE AND STANDARDS

• Own the deployment architecture for each pod inside the client's tenant, and get it approved by their architecture and security functions before build starts.
• Decide what belongs in the harness and what stays a script. Decide where an agent runs, who owns it, how it is versioned and how it retires.
• Keep the architecture decision log current and visible to the client at all times.
• Set the engineering standard across the pods:
  Python conventions, repo layout, CI/CD, secret handling, how client tools are registered behind the connector.

EVALUATION AND QUALITY

• Make sure every workflow has a golden set built from real cases before anyone calls it done. Where the client cannot produce acceptance criteria, say so and hold the workflow.
• Review evaluation sets and traces, not only code. Pick the serving model by evaluation and lock it centrally.
• Own the failure profile that goes to the client, with observed results kept separate from modelled.

PEOPLE

• Review the forward-deployed engineers' work weekly. Unblock them on architecture, Azure and client politics.
• Grow the bench: training plans, pairing, certification paths, so the next pod starts further along than the last.

CLIENT AND COMMERCIAL

• Be the technical counterpart for the client's AI Hub, architecture, security and IT policy people.
• Join phase gates and quarterly operating-model reviews. On site for gates.
• Support presales: shape technical answers in proposals, size pods, sanity-check what we commit to before it is signed.
• Write the monthly market brief that goes to clients: which developments to ignore, test or adopt, with the practical consequence for their stack.

LIMESTONE

• Feed the reusable…