Applied AI Engineer

Overview

Applies and scales AI/ML for the Document AI platform within Trafigura’s Digital Transformation. You will build production-grade Agentic AI/LLM systems to streamline document-heavy workflows and enable data-driven decision making across trading operations. The role combines hands-on development with reliability, monitoring, and collaboration across cross-functional teams. You’ll join a Centre of Excellence focused on agentic AI to impact thousands of users and transform business processes.

Responsibilities

• Develop and maintain Python-based AI applications
• Build and maintain document workflows using LLMs and classical NLP
• Debug model performance issues and optimize system reliability
• Write tests and implement monitoring for AI/ML systems
• Rapidly prototype features, evaluate them, and deploy to production
• Participate in code reviews, technical design discussions, and architecture decisions
• Communicate requirements effectively to technical and non-technical stakeholders
• examples projects you might own in the first 6 months include: develop a specialised agent to automate complex commodity trading workflow; develop an agent to operate within human workflows with UI integrations and durable executions; develop a new AI product to extract key insights from firmwide market intelligence emails for front and middle office

Key requirements

• 5-8+ years of experience building production AI/ML systems
• Modern Python proficiency with Pydantic, FastAPI, asyncio, type safety
• Experience with modern AI frameworks (Pydantic AI favored)
• Production experience with human-in-the-loop systems and AI monitoring/observability
• Strong AI/ML evaluation framework fundamentals
• Experience building agentic systems including MCP, tool calling, memory systems, vector stores, guardrails
• Deep understanding of software design principles (microservices, event-driven, DDD, OOP, TDD)
• Deep understanding of SDLC (CI/CD, IaC, Containerisation)
• Practical cloud engineering experience (AWS preferred)

• Self-directed and autonomous
• Pragmatic about technology choices
• Strong focus on AI system reliability
• Python
• Pydantic
• FastAPI

…