AI Engineer

At Keystone AI , we are seeking an experienced AI Engineer to optimize AI solutions that drive our digital transformation services. Project / Role Design, develop, and maintain AI agents for a production-grade LLM and deterministic Python-based tax pipeline. Build planning and rollover agents using modern agentic workflow patterns. Integrate AI solutions with PwC applications through REST APIs and Model Context Protocol (MCP) tools. Develop agent-to-agent (A2A) compatible components for a graph-based orchestration platform built with LangGraph. Implement scalable, stateful multi-agent workflows with durable execution and human-in-the-loop capabilities. Build reliable, secure, and high-performance backend services using modern Python technologies. Collaborate with cross-functional teams to deliver enterprise AI solutions. Ensure code quality through automated testing, code reviews, and engineering best practices. Requirements Proven experience as a Software Engineer with 1+ year of hands-on experience developing Agentic AI workflows. Strong Python skills, including asyncio, FastAPI, Pydantic, and automated testing with pytest / pytest-asyncio. Practical experience building LLM-based agents, including tool calling, structured outputs, multi-agent architectures, and cost/latency optimization using frameworks such as Claude Agent SDK or similar. Experience with LangGraph, including graph-based orchestration, stateful workflows, human-in-the-loop execution, durable checkpointing, and Agent-to-Agent (A2A) communication. Hands-on experience with Model Context Protocol (MCP) for building and integrating agent tools. Experience integrating enterprise applications through REST APIs using OAuth2 / Azure AD SSO. Knowledge of Redis / Redis Stack, Docker, and modern software engineering best practices. Professional English and strong collaboration skills. Nice to have: Experience with observability platforms such as OpenTelemetry, Datadog, or Langfuse. Familiarity with Azure services (Blob Storage, Managed Search). Experience with LLM evaluation frameworks (LLM-as-a-Judge). Knowledge of specification-driven development. Experience in tax or accounting domains. What do we offer? We focus on long-term relationships based on fair principles and reliability Co-financing of the Multisport card and Medicover private healthcare Modern office available Team bonding activities, internal courses, conferences, certifications