AI Engineer – LLMs, NLP & Market Intelligence

AI Engineer - LLMs, NLP & Market Intelligence Location: London – hybrid, 2+ days a week in our Vauxhall office Employment type: Full-time, permanent Experience: Typically 2–4 years Build AI systems that interpret global markets in real time Permutable is building market intelligence infrastructure that helps financial institutions understand what is happening across global markets – and what is driving it.

Our technology processes large volumes of multilingual news, economic developments, market narratives and geopolitical information and turns them into structured, explainable intelligence for institutional investors, banks, asset managers and trading teams. We are looking for an exceptional AI Engineer to join our London engineering team and help build the next generation of our NLP, LLM and market intelligence systems.

This is a hands-on engineering role for someone who wants considerably more ownership than they are likely to get inside a large technology company, bank or established AI business. You will work directly with experienced engineers, data scientists and our founder, taking problems from experimentation through to production. Your models will not sit in notebooks. They will become part of live systems.

What you’ll work on You will help design and build systems across: Large language models and multi-model architectures Natural language processing and information extraction Agentic and automated research workflows Retrieval, embeddings and semantic search Narrative detection and clustering Multilingual text intelligence Large-scale data and model pipelines Model evaluation, observability and monitoring Production ML infrastructure on AWS The problems are often open-ended.

You might be evaluating how reliably different models identify changes in a market narrative, improving entity resolution across millions of documents, designing an agentic workflow for automated market analysis or reducing the latency of a production intelligence pipeline.

What you’ll do Build production AI systems Design, build and deploy LLM and NLP pipelines that operate reliably at production scale. Take models from experimentation through evaluation, deployment, monitoring and continuous improvement. Develop and evaluate models Build and optimise models in Python using modern machine learning and NLP techniques. Experiment with transformers, embeddings, retrieval systems, fine-tuning and different LLM architectures. Develop rigorous evaluation frameworks rather than relying solely on headline benchmark performance.

Work with proprietary datasets Train and evaluate models against Permutable’s large-scale historical and real-time datasets. Carry out detailed error analysis and use what you find to improve model and system performance. Build reliable ML infrastructure Develop and maintain data and machine learning workflows using technologies such as Apache Airflow. Help improve CI/CD, automated testing, monitoring and reproducibility across our ML stack. Engineer on AWS Build and improve cloud infrastructure using services including S3, ECS/EKS, Lambda and Redshift.

Automate infrastructure and deployment through tools such as GitHub Actions and Pulumi. Own what you build Take responsibility for systems beyond the initial model or prototype. You will be expected to understand how your work behaves in production, investigate failures and improve it over time. Help shape the product Work closely with engineering, data science, market analysts and leadership. We are a small team, so good ideas can move quickly from a conversation to an experiment and into production.

What we’re looking for You will probably have around 2-4 years of professional software engineering, machine learning or AI engineering experience, although we care more about the quality of your experience than the exact number of years.

You should have

Excellent Python skills. Strong software engineering fundamentals. Experience building or deploying machine learning systems beyond experimentation. Practical experience working with modern NLP, transformers or LLMs. A good understanding of machine learning evaluation and experimentation. Experience with APIs, data pipelines and production systems. Familiarity with Docker and modern CI/CD practices. Strong analytical and problem-solving ability. The ability to work independently on ambiguous technical problems.

High standards for reliability, maintainability and technical quality. The confidence to challenge assumptions and contribute ideas. A genuine interest in rapidly evolving AI technology. A strong academic foundation in computer science, engineering, mathematics, physics, machine learning or another quantitative discipline is useful, but we care most about what you can build and how you think. What would make you stand out We would particularly like to meet engineers who have: Built LLM or NLP systems that reached real users.

Worked with transformer architectures or fine-tuned language models. Designed evaluation frameworks for generative AI systems. Built retrieval, RAG, embedding or semantic-search systems. Worked with large, noisy or multilingual text datasets. Built high-throughput or low-latency data pipelines. Experience operating machine learning models in production. Strong understanding of the trade-offs between model quality, latency and cost. Built technically ambitious side projects or contributed to open source.

Published research or completed substantial postgraduate research in ML, NLP or related areas. We are much more interested in what you built, why you made particular technical decisions and what you learned when things did not work than in collecting technology keywords.

Useful experience Experience with some of the following would be valuable, but we do not expect you to know everything: PyTorch, TensorFlow or JAX Hugging Face LLM APIs and open-source models RAG and vector databases Apache Airflow AWS S3 ECS / EKS Redshift Docker Kubernetes GitHub Actions Pulumi or Terraform SQL Model monitoring and observability Distributed processing Financial markets, economics or commodities Financial-market experience is not required. Curiosity about how markets, economics and global events interact is more important.

Why Permutable? Build systems that move into production fast You will work on live AI infrastructure rather than internal prototypes or proof-of-concept projects. Your work can move into production quickly and directly influence the quality of our products. More ownership, earlier We deliberately keep teams small. Strong engineers can take responsibility for important technical problems without waiting years to be given ownership. Work across the full AI stack You will have exposure to models, data, infrastructure, evaluation, deployment and product.

For someone early in their career, that creates an unusually steep technical learning curve. Work close to the problem Engineers work directly with data scientists, market analysts and leadership rather than receiving requirements through several layers of management. You will understand not only what you are building, but why. Influence technical direction We expect engineers to propose ideas, challenge existing approaches and run experiments. If you find a better way of solving a problem, we want to hear it.

Grow as the company grows We are building an ambitious technology company in London. As the platform expands, there will be opportunities for strong engineers to take ownership of increasingly significant systems and technical areas. You’ll probably enjoy this role if you… Want to build rather than coordinate. Like difficult technical problems without obvious answers. Want your work to reach production. Learn new technologies quickly. Are comfortable moving between ML research and engineering. Prefer ownership to narrowly defined responsibilities.

Want to work around highly capable people in a small team. Find the intersection of AI, data and global markets intellectually interesting. A startup will not give you the structure or predictability of a large corporate engineering organisation. In return, you will have far greater visibility into the whole system, considerably more responsibility and the opportunity to influence what gets built.

How to apply Please send us your CV and, where possible, something that gives us a better sense of you as an engineer - for example: GitHub or an open-source contribution A technical project Research or a dissertation A system you have built professionally A short explanation of a particularly difficult engineering problem you have solved We look forward to hearing from you!