Software Engineer (LLM Engineering)

Isomorphic Labs; City of Westminster; On-site; GBP 120,000 - 180,000; Full time

Job summary

Isomorphic Labs seeks a senior ML infrastructure professional to architect LLM-integrated systems with a strong emphasis on security, scalability, and user experience. You will build tools for observability, governance, and cost management, while enhancing agentic tooling and RAG workflows.

Collaboration with researchers to translate scientific needs into AI specifications is essential. The role focuses on production-ready software, modern LLM serving stacks, and cloud-based ML infrastructure

Qualifications

• Strong coding skills in Python for production-grade, maintainable systems.
• Experience with LLM serving stacks such as vLLM.
• Familiarity with GCP and cloud architectures for ML workloads.

Responsibilities

• Architect LLM-integrated systems with security, scalability, and UX in mind.
• Build telemetry, observability, governance, and cost management tooling.
• Develop agentic infrastructure including tool-use and RAG capabilities.
• Collaborate with researchers to translate scientific needs into AI specs.

Skills

Python

Stakeholder management

Communication

First-principles thinking

ML literacy

Tools

vLLM

Google Cloud (GCP)

Job description

As a member of the LLM Engineering team, you will work on a subset of the following:

• System Architecture: Architect LLM-integrated systems by incorporating scalability, security, and user experience considerations from the earliest stages of development.
• Operational Excellence: Enable telemetry, observability, governance and cost management. Build platform components and tools for routing and making optimal use of various models.
• LLM & Agentic Infrastructure: Implement advanced context management, tool-use (MCP), skill tooling, and RAG to enhance research workflows and help scale our agentic infrastructure.
• Evaluation Frameworks: Build rigorous testing for probabilistic systems to ensure model reliability and safety. Build and maintain evals for different tasks.
• Agentic Tooling: Develop the internal frameworks and IDE integrations to leverage agentic coding and agentic workflows at scale. Contribute to shaping best practices and providing a great user experience.
• Refinement & Training: Support fine-tuning and RL efforts in collaboration with AI researchers to optimise models for complex biological and chemical data.
• Strategic Collaboration and Solutioning: Partner with researchers to translate scientific needs into technical AI specifications. Partner with other parts of the business to unlock LLM-based use-cases. Software Engineering: Strong coding skills (Python) with a focus on production-grade, maintainable systems.
• System Design: Ability to architect and maintain (IaC) complex systems across multiple verticals (Security, UX, and Scalability) in the cloud.
• LLM Internals: Deep understanding of how models are trained, how they work internally, and their inherent limitations.
• LLM Serving stack: Experience with the LLM serving stack (e.g. vLLM) for open weight models for bringing the latest models to internal users.
• ML Literacy: Experience evaluating probabilistic ML systems and managing model "tool use", contexts and loops.
• First-Principles Thinking: A hype-detached approach to solving problems and selecting the right tech stack.
• Communication: Excellent stakeholder management and the ability to explain technical risks to non-experts., Applied LLM Experience: Prior success building and scaling systems centered around Large Language Models.
• ML Infrastructure: Experience setting up the underlying hardware or orchestration for ML systems.
• GCP: Prior experience building and deploying systems on Google Cloud (GCP).
• Scientific Domain: Interest or experience in biology, chemistry, or drug discovery; an interest in applying AI to scientific discovery.
• Fast-Moving Tooling: Familiarity with the latest in agentic tooling and developer frameworks. Isomorphic Labs (IsoLabs) was launched in 2021 to advance human health by building on and beyond the Nobel-winning AlphaFold system. Since then, our interdisciplinary team of drug discovery experts and machine learning specialists has built powerful new predictive and generative AI models that accelerate scientific discovery at digital speed.