Python AI agent engineer in finance

We are looking for a hands-on Lead Full-Stack AI Agent Engineer to drive the technical development of an enterprise-grade multi-agent AI platform and a suite of analytical AI agents for the Finance & Accounting domain. In this role, you will lead the engineering of a production-ready agentic platform built on Microsoft Azure and Microsoft Foundry, combining LLM-based reasoning, deterministic business logic, and statistical anomaly detection to deliver reliable, secure, and scalable AI-driven analytical solutions.

• Lead a dedicated AI engineering team, provide hands-on technical guidance, conduct code reviews, establish engineering standards, and develop reusable agent templates and architectural patterns
• Design and implement the core agentic architecture, including the Master Orchestrator, Agent Registry, multi-LLM routing (Claude and Azure OpenAI), and multi-level caching mechanisms (Redis and prompt caching)
• Develop shared context and memory services to manage domain glossaries, business rules, analytical thresholds, and persistent user feedback.
• Implement mechanisms that allow agents to retain user confirmations, dismissals, and false-positive suppression rules
• Build specialized analytical and control agents that work with structured financial and ERP data.
• Combine deterministic business rules, Python-based statistical anomaly detection, and LLM-generated explanations supported by data references and citations
• Implement secure identity propagation using Microsoft Entra ID and On-Behalf-Of (OBO) flows, Role-Based Access Control (RBAC), and row-level security.
• Ensure data freshness, traceability, and reliable citation mechanisms to prevent unsupported AI-generated insights and unauthorized data access
• Implement operational and business KPI monitoring, build telemetry dashboards, lead production releases, and support hypercare activities and knowledge transfer to the client's IT team

• Advanced experience with Python 3.11+, asyncio, FastAPI, Pydantic v2, pytest, strict typing, and clean, modular software architecture
• Hands-on experience designing and implementing multi-agent AI systems using Microsoft Foundry (Azure AI Agent Service), Semantic Kernel, LangGraph, or AutoGen.
• Understanding of orchestrator–subagent patterns, MCP (Model Context Protocol), and structured tool calling
• Production experience with Redis for session management and semantic caching, as well as PostgreSQL, Cosmos DB, or similar databases for persistent agent memory, audit logs, and user feedback
• Practical experience with Microsoft Entra ID (Azure AD), OAuth 2.0 On-Behalf-Of (OBO) flows, RBAC, and Azure Key Vault
• Experience with OpenTelemetry, Azure Application Insights, Langfuse or Microsoft Foundry tracing, prompt caching, and LLM token usage and cost optimization
• Будет плюсом: Experience developing AI-driven solutions for Finance, Accounting, Audit, or ERP analytics; Experience integrating AI applications with Microsoft Fabric or enterprise SQL data warehouses; Experience designing enterprise-grade AI platforms with a focus on scalability, security, governance, and performance

• Условий в вакансии нет