Machine Learning Engineer

Oscar Technology; Greater London; On-site; GBP 95,000 - 120,000; Full time

Job summary

Oscar Technology in London is seeking a Machine Learning Engineer to join a production ML stack, building LLM-powered features and robust evaluation systems for financial data workflows. You’ll design, implement and monitor automated pipelines that analysts rely on to extract insights from complex documents.

You’ll own the ML layer from prototype through production, work in Python, and collaborate with a UK core team to ship reliable, scalable services in a regulated multi-tenant SaaS

Qualifications

• 5+ years in machine learning or software engineering with production ML experience.
• Hands-on experience shipping LLM-based or NLP systems to real users, including RAG, agents and tool use.
• Rigorous evaluation: define what good means, build test sets and measure it.
• Strong grounding in ML stack: PyTorch, scikit-learn, Hugging Face and vector databases.
• Experience deploying and monitoring models on cloud infrastructure, ideally AWS, with Git, CI/CD and containerisation.

Responsibilities

• Design, build and ship LLM-powered features and multi-step agent workflows in Python, from prototype to production.
• Build document intelligence pipelines to extract and reason over financial documents.
• Develop retrieval-augmented generation systems with traceability to sources.
• Create evaluation frameworks and datasets with automated tests and human review loops.

Skills

Production ML experience

Strong Python

LLM/NLP systems

Model evaluation

Cloud infrastructure (AWS)

Education

CS/Engineering/Math degree

Postgraduate degree valued

Tools

PyTorch

scikit-learn

Hugging Face

Vector databases

Git

CI/CD

Containerisation

AWS

Job description

Machine Learning Engineer

Location: London

Salary: £95,000-120,000

Contract: Permanent

Working pattern: On site, 3 days a week

A new software venture built inside an established global provider of analytical and research services to the private capital industry is hiring Machine Learning Engineers to build the models, agents and evaluation systems that investment professionals rely on to analyse data, work with documents and automate research. You'll join the UK core team of around thirty people, working in Python on production ML and LLM systems that sit behind data-dense enterprise products.

THE ROLE

You’ll report to the Head of Engineering and own the machine learning layer of the platform, from the first prototype through to a monitored production service. The work combines applied LLM engineering with the rigour that high-stakes financial workflows demand.

• The hard part is reliability and measurement: a model that is right most of the time is not enough when analysts act on the output, so every system needs to be evaluated, observable and safe to fail.
• Part of an engineering cohort being hired across machine learning, front-end and back-end.
• Based with the UK core team of around thirty people in Kings Cross, alongside colleagues in London and India.

WHAT YOU'LL BE DOING

Your remit spans the models and agents, the data and retrieval layer, the evaluation framework and the production platform.

• Design, build and ship LLM-powered features and multi-step agent workflows in Python, from prototype to production.
• Build document intelligence pipelines that extract, structure and reason over long, messy financial documents such as filings, decks, models and data rooms.
• Develop retrieval‑augmented generation systems, including chunking, embeddings, vector search, re‑ranking and citation, so that every answer can be traced to its source.
• Create evaluation frameworks and datasets, with automated regression tests, human review loops and metrics for accuracy, that catch mistakes early and stop the same failures recurring.
• Implement guardrails, confidence signalling and safe failure behaviour, and work with front‑end colleagues to expose approvals, tool calls and partial progress to users.
• Fine‑tune, prompt‑engineer and benchmark models, and make considered decisions between hosted foundation models, open‑weight models and classical ML on cost, latency and quality.
• Build and operate ML services and APIs with authentication, role‑based access and tenant isolation, so client data is never mixed across tenants.
• Own MLOps in production: experiment tracking, model and prompt versioning, monitoring, drift detection, cost and latency management, and incident response.
• Write unit, integration and evaluation tests, and ship through Jenkins and SonarQube pipelines to AWS.
• Work with back‑end and front‑end engineers and with domain experts in private capital to turn real analyst workflows into reliable products.

WHAT WE'RE LOOKING FOR

• 5+ years in machine learning or software engineering, with production ML experience and strong Python.
• Hands‑on experience shipping LLM‑based or NLP systems to real users, including RAG, agents and tool use, and the ability to explain the design and trade‑off decisions behind them.
• A rigorous approach to evaluation: you can define what "good" means for a model, build the test set and measure it.
• Strong grounding in the ML stack, such as PyTorch, scikit‑learn, Hugging Face and vector databases, alongside sound software engineering practice.
• Experience deploying and monitoring models on cloud infrastructure, ideally AWS, with Git, CI/CD and containerisation.
• Experience building SaaS or multi‑tenant products, with a working understanding of data security, access control and privacy.
• Daily use of AI coding tools such as Cursor or Claude Code, and a considered view of where they help.
• A degree in computer science, engineering, mathematics, statistics or a related field, or equivalent experience. A postgraduate degree is valued but not required.

NICE TO HAVE

• Experience building agentic products, including orchestration, memory, tool calling and human‑in‑the‑loop controls.
• Document understanding, OCR and table extraction on complex financial documents.
• Fine‑tuning or distillation of language models, and inference optimisation.
• Experience with financial, investment or other regulated‑industry data.
• Experience with LLM observability and evaluation tooling.

THE REALITY

• Autonomous systems are being built inside high‑stakes live businesses, where decisions carry real consequences and waiting for everyone to agree is not always an option.
• The systems will sometimes get things wrong; in this phase the London team needs to create the evaluations, visibilities and safeguards that catch mistakes early and stop the same failures recurring.
• The goal is to build the foundations of an enterprise‑grade technology platform serving a multi‑billion dollar serviceable market, and to share in that growth.

IN OUR CONVERSATIONS

Come ready to discuss three pieces of work:

• A production ML or LLM system you shipped: the architecture, data and cost decisions.
• An evaluation approach you designed to measure or improve quality, and what it revealed.
• A failure or regression you found in production and the improvement you made to prevent it recurring.

LOCATION & WORKING PATTERN

• Kings Cross, London. In office 3 days a week.
• Colleagues based in London and India.
• Visa sponsorship is not available for this role.
• The interview process is intended to complete within three weeks, and adjustments to interviews can be arranged on request.