Dot Net AI

Role Summary
We're hiring a hands-on senior engineer to do two connected things for client
engagements: (1) architect and build an Azure-native AI Gateway - functionally equivalent
to a LiteLLM + Langfuse stack (multi-model routing, rate limiting, semantic caching,

token/cost metering, tracing) - and (2) design and deliver multiple Agentic AI proof-of-
concept scenarios in C# using the Microsoft Agent Framework. This is an individual-
contributor role with direct client exposure, not a platform-team or people-management

position.
Top 5 Responsibilities
1. Architect and implement the AI Gateway. Design and build an Azure-native LLM
gateway in C# providing unified ingress across multiple model providers (Azure
OpenAI, Azure AI Foundry models, and others) - covering intelligent routing,
fallback/load balancing, rate limiting and token-quota enforcement, semantic
caching, and centralized API key/secret management. Reference the LiteLLM
feature set as the functional bar to hit.
2. Build the observability and governance layer. Implement request/response
tracing, prompt/completion logging, token and cost metering, and latency
dashboards - the Langfuse-equivalent half of the stack - using Azure Monitor,
Application Insights, OpenTelemetry, and APIM's native LLM logging/token-metric
policies (or a self-hosted Langfuse instance where warranted).
3. Deliver multiple Agentic AI PoC scenarios. Using C# and the Microsoft Agent
Framework, build a portfolio of distinct agentic patterns (single-agent tool use,
multi-agent orchestration, human-in-the-loop workflows, RAG-grounded agents)

mapped to real client business use cases - not one deep PoC, but several breadth-
covering scenarios that demonstrate different capabilities.

4. Own technical delivery on client engagements end-to-end. Run architecture
proposals, hands-on build, live demos, and production-readiness assessments

2
directly with client technical stakeholders; translate ambiguous business asks into
scoped, demoable agentic scenarios.
5. Package the work as reusable engineering assets. Turn the gateway and agent
scenarios into templates, SDKs, or IaaC that can be re-deployed across
environments rather than rebuilt from scratch each time - even though this is a
client-facing role, the artifacts should outlive any single client environment.
Top 5 Required Skills / Background
1. Deep C#/.NET engineering background (12-15+ years). Production-grade
API/service development, async patterns, dependency injection, testing discipline,
and comfort operating without a large surrounding platform team.
2. Hands-on Azure platform expertise. Azure API Management (including
GenAI/LLM-specific policies: token-limit, token-metric, semantic-caching), Azure
OpenAI / Azure AI Foundry, Azure Monitor / Application Insights, Key Vault, and
either Azure Functions/App Service or AKS for hosting gateway services.
3. Working knowledge of LLM gateway and observability platforms. Familiarity with
LiteLLM, Langfuse, or comparable tools (Portkey, Kong AI Gateway) - not to operate
them directly, but to translate their proven patterns (routing, fallback, semantic
caching, token metering, tracing) into an Azure/C# implementation.
4. Microsoft Agent Framework experience, or a direct path into it. Hands-on work
with MAF, or strong prior experience in Semantic Kernel and/or AutoGen (MAF's
predecessors) - multi-agent orchestration, tool/function calling, thread/state
management, and ideally MCP/A2A protocol exposure.
5. Client-facing solution engineering track record. Demonstrated ability to
architect, prototype, and present technical PoCs directly to enterprise client
stakeholders, defend design trade-offs live, and adapt scope under engagement
time pressure.