Ai Engineer
Posted Updated
Description
Key Responsibilities
- Own agent, scoring and API code end to end; ship in two-week increments against written specs
- Define the feature contract with the data engineer; tests, CI, runbooks
- Nightly targeting agent (Foundry hosted agent + Routine): typed tools over Fabric feature tables,constraint-based selection, reason per row, list write-back
- Scoring: configurable weights, then a learned ranker (LightGBM) trained on call outcomes
- Backend API (Python, FastAPI) for the Teams/Outlook app: feeds, outcomes, follow-up monitor,
- Microsoft Graph on-behalf-of: mailbox reply scan, Outlook draft creation; Entra app registration,
- Salesforce adapter (MCP or Connect API) for reads and idempotent writes
- Structured-output LLM calls, validators, prompt versioning, evaluations, tracing, agent identity and guardrails.
Profile
- Must have: Production Python/FastAPI; Azure OpenAI / Foundry Agent Service; Microsoft Agent Framework or equivalent; Graph OBO and Entra; one shipped Salesforce integration; practical ranking ML
- Plus: Teams Toolkit, Container Apps, Service Bus, Azure SQL/Cosmos, Bicep, GitHub Actions; AIassisted development