We're looking for a Senior .NET Engineer who builds production agentic systems and works the AI-native way: coding agents, specs, and evals are part of the daily workflow. You'll design scoped AI agents and the tools and MCP servers behind them, measure their quality, and ship them as reliable, observable and secure .NET services. Responsibilities Agent and Tool Development: Build, test, and maintain scoped AI agents (e.g., Resource Optimization, Metadata, Contract Management). Create custom agent tools, system prompts, and structured JSON output parsers MCP and API Engineering: Implement read-only data access tools on MCP servers to bridge agents with the data access layer safely. Apply least-privilege access and OAuth2 service accounts Observability & Health: Implement standardized /health and /ready probes, structured JSON logging, W3C trace propagation (OpenTelemetry) and real-time agent execution status monitoring. Track latency, cost and token usage per run Evaluations: Test and evaluate agentic workflow performance and quality, and tool-call trajectory. Improve RAG Retrieval and Generation precision and quality; work with evaluation metrics; define and use golden datasets AI-native delivery: Use coding agents (GitHub Copilot, Claude Code) and spec-driven development in daily work. Review and verify generated code for correctness, security and maintainability. Codify what works as reusable agents, skills, prompts and workflows for the team Collaborative Development: Co-locate unit and integration tests with capability modules, take part in peer code reviews, and share AI-engineering practices with teammates Requirements Experience: 3+ years of production .NET (C#) development, and 1+ year building LLM integrations and multi-agent workflows in production Agent frameworks and tool calling: Hands-on experience building agents with at least one framework, ideally Microsoft Agent Framework or Semantic Kernel (or LangChain/LangGraph, Google ADK, Strands or equivalent). Solid tool/function calling Agentic RAG and context engineering: Hands-on work on retrieval, agent memory, planning and routing. Practical context-engineering skills, including managing context across multi-step tasks Multi-agent solutions: Orchestration patterns and guardrails LLM APIs: Experience with at least one stack (Azure OpenAI / AI Foundry, AWS Bedrock, Vertex AI or LiteLLM) and Microsoft.Extensions.AI MCP and integration: Building MCP servers or tools, ideally with the C# MCP SDK on ASP.NET Core. RESTful API design, OAuth2 service accounts, event-driven communication, microservices Observability and evaluation: OpenTelemetry and structured logging for services. Basic LLM evaluation (metrics, golden datasets, tool-call checks) Data: SQL (e.g. SQL Server, PostgreSQL) and at least one NoSQL store (e.g. Cosmos DB, MongoDB, Redis) Security: OWASP Top 10 for LLM, PII handling, input/output validation, MCP security AI-assisted engineering: Daily use of at least one coding agent (GitHub Copilot or Claude Code), and a habit of reviewing and verifying generated code for correctness, security and maintainability Nice to have Spec-driven development (OpenAPI/JSON Schema as the source of truth), and codifying what works as reusable agents, skills, prompts and workflows for the team Agentic Context Engineering (ACE) Azure/AWS (App Service/AKS, Azure DevOps), Docker/Kubernetes, CI/CD Vector databases (pgvector, Azure AI Search), LLM-as-judge and red-teaming experience
