Applied AI Engineer

Overview

In this AI Engineer role you will shape and maintain an agentic AI platform and Python SDKs to empower research and business teams. You’ll deliver production-grade agentic workflows, set standard ways of working, and embed with domains to uncover new AI opportunities. You’ll balance platform development with hands-on delivery and continuous improvement of evaluation and observability. Join a pragmatic, cross-functional team building scalable, high-impact tools for intelligent systems at G-Research.

Pay / Benefits

• highly competitive compensation
• lunch provided
• 30 days annual leave
• 9% company pension contributions
• comprehensive healthcare and life assurance
• cycle-to-work scheme

Responsibilities

• Build and evolve the agentic AI platform, abstractions, and orchestration patterns
• Develop Python SDKs with firm-specific abstractions for agent scaffolding, tool integration, context management, evaluation and deployment
• Integrate open-source tooling (LangGraph, Pydantic AI, etc.) with firm abstractions
• Define and promote best practices for agent development including evaluation standards and testing approaches
• Lead by example on agentic evaluations by building and operating evaluation pipelines (LangSmith, Langfuse)
• Deliver production agentic solutions for internal stakeholders and validate abstractions in real-world conditions
• Embed with business teams to understand domains and deliver tailored AI solutions
• Apply context engineering to optimize information retrieval, structure, and usage
• Fine-tune and optimize models for domain-specific accuracy, latency, and cost
• Integrate with existing stacks (C#, C++, JVM) ensuring clear APIs, monitoring, and CI/CD
• Upskill engineers through pair-programming, workshops, SDK docs, and playbooks
• Stay updated on the LLM ecosystem and feed lessons back into the platform

Key requirements

• Hands-on experience building LLM applications with LangGraph/LangChain, Pydantic AI, FastAPI, MCPs, and RAG (pgvector, Pinecone, Qdrant, Milvus)
• Strong understanding of context engineering (retrieval strategies, context windows, dynamic prompts, information routing)
• Experience designing complex agentic workflows (multi-step planning, tool use, self-correction, multi-agent patterns)
• Solid understanding of RAG patterns, prompt engineering, and safe deployment
• Experience with agentic evaluation frameworks (LangSmith, Langfuse) for measuring accuracy, latency, cost, and regressions
• Proven Python production expertise with async patterns, typing, testing; building platform-level software
• Ability to integrate with heterogeneous stacks (REST/gRPC, SQL/NoSQL) and automate deployment (Git, Docker, Kubernetes)
• Strong communication skills to translate requirements and articulate trade-offs

• strong communication with technical and non-technical audiences
• stakeholder management
• ability to work across green-field and production environments
• LangGraph
• LangChain
• Pydantic AI

…