You will build evaluation, experimentation, and observability pipelines using Python and platforms like Braintrust, LangSmith, Arize, and Weights & Biases, plus design LLM-as-judge and human-in-the-loop scoring. The role focuses on AI-native consulting for clients like PayPal and adidas, including production monitoring, quality signal detection, and cost attribution. You'll also wire evaluation gates into CI/CD pipelines for the AI Engineering team.

Build REST and GraphQL APIs and MCP servers in Python (FastAPI, Flask) to expose ML models from Databricks to internal apps, CRM platforms, and AI agents at RBC. Work within the Data Science team to architect the serving layer, handling contract design, authentication, and documentation.

You'll work with Microsoft Fabric, PySpark, Python, T-SQL, DAX, Infrastructure as Code, Git, and Azure DevOps. You'll design, build, and maintain data pipelines, lakehouses, and data warehouses within Fabric to drive business intelligence and analytics.

Wise★3.5 (42)· London

You will build scalable systems for business onboarding and verification using Java or JVM languages, Spring framework, JDBC, and relational/non-relational databases with CI/CD pipelines and distributed systems. This engineering team owns end-to-end projects to automate verification, create flexible onboarding journeys, and manage compliance risks for Wise’s money-moving platform across Europe and the UK.

Uses SQL, Python or scripting, BI tools like PowerBI/Looker/Tableau, spreadsheets, and Git to investigate fraud and build dashboards. The role sits in fraud analytics at Moniepoint, an all-in-one financial services platform powering POS transactions and banking across Nigeria and emerging markets.

DataCamp★3.8 (4)· London

You will work with agentic AI systems like Claude Code, Cursor, or Replit to build AI workflows that draft, grade, and quality-check assessment content. You will own the design and production of DataCamp's assessment and certification content, collaborating with subject-matter experts and teams across curriculum, engineering, product, and marketing. This is an individual contributor role.

Design, build, and optimize analytics solutions using SQL, DBT, Power BI (DAX, RLS), GitHub, and cloud platforms (Snowflake, BigQuery, Databricks, Fabric). You will drive data-driven marketing insights for enterprise clients, architecting data models and leading test-and-learn programs. The role also involves mentoring junior team members.

This role involves building ML models for low-data drug discovery using active learning, meta-learning, fine-tuning, uncertainty estimation, and multimodal modeling, with tools like PyTorch and JAX. The scientist will design data acquisition strategies and closed-loop workflows for Lila’s AI-driven scientific discovery systems, collaborating with drug discovery and computational teams.

LexisNexis Legal & Professional★3.6 (1k)· London

You will drive AI ideation and integrate LLMs, NLP, and machine learning into Practical Guidance editorial workflows. This global role partners with commercial, product, and engineering teams to deploy scalable content-optimization solutions across North America, UK, and Asia-Pacific.

PlayStation★3.8 (164)· London

This role involves working with TensorFlow or PyTorch, Python, and libraries for computer vision and graphics. As part of the Visual Computing Group (VCG), you will develop generative models to advance game rendering and live video streaming through neural networks and machine learning R&D.

Work with Python, PyTorch, JAX, CUDA, Docker, Kubernetes, Slurm, Ray, and Linux on GPU computing. The Molecular Intelligence team builds ML and physics-based infrastructure for drug discovery, including biophysics, chemistry, cofolding models, simulation workflows, and agent-usable scientific pipelines.

Train cofolding models for protein-ligand drug discovery using PyTorch, JAX, contrastive learning, and structure-aware methods like Boltz, AlphaFold, and equivariant GNNs. The role develops models that reason over proteins, ligands, and experimental DEL data to improve AI-driven discovery decisions.

Relay Technologies★2.0 (9)· London

At Relay Technologies you will build and own time-series forecasting models for parcel volume, inbound international volume, and parcel size/weight, using Python and ML frameworks. You will work on the Network squad’s demand-forecasting engine that predicts parcel movement across sortation, middle-mile, and last-mile operations, directly affecting cost per parcel. This is a hands-on individual-contributor role, serving as the technical anchor for Demand Forecasting, collaborating with a Senior Data Scientist, ML Engineer, and Analyst.

You will work with Python, SQL, BM25, vector search, reranking, RAG, and agentic systems to own and evolve Corestream, Kharon’s risk intelligence feed that surfaces entity-level insights and risk narratives for customers in global security, sanctions, and financial crimes.

Python, SQL, BM25, vector search, reranking, and RAG pipelines powering agentic systems. You will own and evolve Corestream, Kharon’s risk intelligence feed, redesigning semantic search and narrative generation for global-security insights.

Build and deploy AI agents for geopolitical and financial risk intelligence, working with Python, SQL, RAG pipelines, BM25, vector/hybrid retrieval, APIs, Docker, Kubernetes, GitHub Actions, and AWS services like RDS, EKS, and S3. Kharon provides risk-analysis tools for global financial institutions and multinationals, covering sanctions, export controls, and threat identification.

Work with Python, SQL, graph databases, graph ML, Docker, Kubernetes, and LLMs. Own and evolve Kharon's risk propagation engine that analyzes entities in a knowledge graph to surface risk signals for global security clients.

HarbourVest★3.1 (16)· London

This role involves productionizing Data Science and AI prototypes using Python, containerized or cloud-based delivery, Snowflake, Power BI, Git, and CI/CD pipelines. You will design analytics workflows and support legacy solutions, partnering with global teams to bridge raw data and analytics-ready assets at HarbourVest, a private markets investment firm.

Moody's★3.6 (589)· London

Lead validation of credit rating models and AI model risk management at Moody's, working with R, Python, MATLAB, SQL, and preferably C++. The MRG Quantitative Review team independently validates quantitative models and scorecards supporting credit ratings, with a focus on Generative and Agentic AI risks. Requires 10+ years in model development, validation, and risk analytics.

MAREX★3.2 (6)· London

You will work with Power BI, alongside a Data Lakehouse architecture and a central data catalogue, to analyze and validate structured and unstructured risk data. You will act as the bridge between the Risk function and Technology to support data-led risk management, oversight, monitoring and reporting across the firm.

Chubb Insurance★3.7 (1.3k)· London

You will use Python, Hugging Face Transformers, LangChain, LlamaIndex, PyTorch, TensorFlow, Docker, and cloud platforms to build and deploy LLM, RAG, and conversational AI applications. As part of Chubb Insurance's Global Analytics team, you'll design production-ready AI solutions for business value.

You'll work with Java/JVM and Spring on the backend, React/JavaScript/TypeScript on the frontend, plus CI/CD, distributed systems, and relational/non-relational databases. You'll join Wise's Account Integrity team, building fraud-prevention and financial-crime detection systems to protect customer accounts. Wise is a global technology company for moving and managing money.

Tesco★3.5 (25k)· London

You will work with Java, Spring Boot, Kafka, Apache Flink, Apache Beam, NoSQL databases, Redis, and Apache Camel. The Tesco Personalisation team builds real-time, low-latency services for recommendations, offers, and experimentation across web, mobile, and in-store channels using streaming data and ML.

You use Python and supervised machine learning on tabular data to own and build production credit models that decide which small businesses iwocaPay’s B2B Buy Now Pay Later product can fund. You co-own the full credit modelling stack with one other data scientist.

You'll work with Python, SQL, PySpark, and the PyData stack to build machine learning and analytics solutions for McKinsey's clients across industries, translating business challenges into statistical models and deploying them to production. QuantumBlack, AI by McKinsey is a data science and AI consulting firm applying advanced analytics to solve complex client problems. You'll collaborate with Data Scientists, Engineers, Designers, and Product Managers globally while based in London, with exposure to R&D projects and conference presentations.