Senior Machine Learning Engineer

Annual releases gave way to two-week sprints years ago, and the platform now handles over 200 million API calls a day. We started in 1992 and have grown to about 4,100 people. We keep recurring meetings to Tuesdays and Thursdays where we can.

Overview

Our London team is growing and we are hiring a Senior Machine Learning Engineer. This hire will help build out our ML platform: feature store, training pipelines, model registry and deployment. We work hybrid out of London, usually three days in, two days remote.

What you'll do

• Track experiments, artifacts and versions in MLflow or Weights & Biases
• Maintain feature pipelines and a feature store such as Feast or Tecton
• Write clean, tested Python and review teammates' code
• Productionize data science prototypes without rewriting everything
• Fine-tune open-weight models with LoRA when it beats prompting
• Monitor models for drift, data quality issues and latency regressions
• Deploy models behind low-latency APIs with FastAPI, Triton, TorchServe or vLLM
• Develop RAG pipelines with embeddings, vector stores (pgvector, Pinecone, Weaviate) and rerankers

Qualifications

• 5+ years of software engineering experience, with at least two spent on ML systems in production
• Comfortable with SQL and large-scale data processing in Spark or Ray
• Understanding of core ML concepts: overfitting, evaluation metrics, bias in training data

Bonus points

• CUDA or Triton kernel experience
• Distributed training with DeepSpeed or FSDP
• Guardrails, red teaming or safety evaluation for LLM features

What we offer

Compensation: £113,000 - £137,000 per annum

• Healthcare cash plan
• Company pension
• 33 days holiday including bank holidays
• Life assurance (4x salary)
• Death in service benefit
• Flexible working hours
• Enhanced maternity and paternity pay

We may close the vacancy early if we receive a high number of applications.

Unfortunately we cannot offer visa sponsorship for this role.

Start: Immediately.

Demo job created on 11jobs for testing and demonstrations. It is not an actual opening at Canopycast.