Machine Learning Engineer - Python & LLM Systems in London

London, Vauxhall | Hybrid: at least three days in the office each week.

Help build systems that turn unfolding world events into market intelligence. At Permutable, we build AI systems that analyse global news, market and proprietary data to produce market intelligence and systematic signals for financial institutions. Our work connects developments across energy, commodities, economic policy and geopolitics to the assets and markets they influence. Behind it are demanding engineering problems: interpreting unstructured information, evaluating model behaviour and running reliable pipelines as new data arrives.

We’re looking for a Machine Learning Engineer with strong Python skills who wants to take models from experimentation into production. You’ll work closely with our founder and engineering team, contribute to technical decisions and see your work become part of the product. As our systems evolve towards real-time processing and more complex AI workflows, you’ll help decide how we build, evaluate and operate them.

What you’ll do:

• Build production LLM pipelines.
• Develop systems spanning data preparation, model training and fine-tuning, evaluation, deployment and ongoing monitoring.
• Improve model performance using Python and our in-house GPU infrastructure to experiment with model architectures and fine-tuning approaches, assessing quality alongside latency and compute requirements.
• Make evaluation part of development by building repeatable evaluations, investigating failure cases and using the results to guide model and pipeline improvements.
• Orchestrate data and model workflows by building and maintaining Apache Airflow pipelines, with monitoring, retries and clear approaches to failure recovery.
• Deploy and operate AI services using AWS services including Lambda, ECS/EKS, S3 and Redshift, supported by Docker, GitHub Actions and Pulumi.
• Take responsibility for what you build by following your work through deployment, monitoring its behaviour in production and resolving problems with the team.
• Contribute to engineering decisions by bringing ideas, testing assumptions and helping choose practical approaches to the next technical challenge.

Why join Permutable?

• Work on a substantial technical challenge by helping turn large volumes of changing, unstructured information into intelligence financial institutions can use.
• Have a voice in what gets built by working directly with our founder and engineers, contributing to model choices, evaluation methods and system design.
• Develop across the full ML lifecycle, building experience in experimentation, GPU-based training, cloud deployment and production operations.
• See the outcome of your work as your models and pipelines will become part of a live product, with responsibility extending beyond the initial deployment.
• Join a close-knit team with a strong startup culture where we share ideas, help each other solve problems and enjoy spending time together, with regular staff socials, offsites, sports events and office treats.
• Share in what you help build with equity options that give you a stake in Permutable’s long-term growth and success.

Requirements:

• At least two to three years of professional AI/ML engineering experience, or equivalent experience delivering substantial, relevant systems.
• Strong Python skills and software engineering fundamentals, including maintainable code, testing, debugging and version control.
• Hands-on experience building, training or fine-tuning machine learning models, with a clear understanding of how to evaluate their performance.
• Experience taking models or AI pipelines into production and understanding what is needed to keep them working reliably.
• Familiarity with MLOps practices, including CI/CD, containerisation, automated testing and monitoring.
• A grounding in computer science, engineering, machine learning, mathematics or a related discipline, developed through study or practical experience.
• Curiosity, initiative and the ability to work through unfamiliar problems.

Useful additional experience:

• LLMs or NLP, GPU workloads, Apache Airflow, AWS, infrastructure-as-code tools such as Pulumi, or financial-market data.