Senior Machine Learning Engineer

present of AI; Greater London; Hybrid; GBP 90,000 - 120,000; Full time

Job summary

Encord is seeking a hands-on software/ML platform engineer to build and operate production systems serving computer vision and multimodal models. The role focuses on deploying DL models, tuning inference for speed and GPU costs, and owning distributed systems from design through live operation.

Collaboration with researchers and product teams is essential to turn prototypes into reliable features. The position centers on production-grade serving, batching and caching strategies, and scalable

Qualifications

• Hands-on engineering role focused on building and running systems for computer vision and multimodal models in production.
• Experience deploying deep learning models and tuning inference for speed and GPU cost.
• Ownership of distributed systems from design through live operation.

Responsibilities

• Build and operate serving systems for vision and multimodal models in production.
• Deploy and optimize deep learning models and inference latency/cost.
• Own distributed systems end-to-end from design to live operation.
• Collaborate with researchers and product teams to turn prototypes into reliable features.

Tools

Kubernetes

Job description

This is a hands on engineering role focused on building and running the systems that serve computer vision and multimodal models in production. A strong candidate brings years of software engineering plus real experience deploying deep learning models, tuning inference for speed and GPU cost, and owning distributed systems from design through live operation. The person also needs to work well with researchers and product teams, turning prototypes into dependable features.

How it fits the AI era

The work sits at the center of how AI changes annotation: models suggest labels in real time and learn from human corrections, while the engineer builds the serving layer that makes this fast and reliable. Much of the craft stays human, including architecture calls, trade offs between latency, throughput and cost, and judging when a simple baseline beats a clever one. The role is less about training models and more about making them dependable at scale.

How it builds AI experience

The job offers direct exposure to serving transformer based vision and multimodal models, inference optimisation, GPU orchestration and agentic quality control at production scale. Someone building toward this could practise deploying models with batching and caching, working with Kubernetes and cloud GPU workloads, and handling large video or sensor datasets. Experience with computer vision or multimodal deployment helps but is listed as a bonus rather than a requirement.

Worth showing

• Inference optimisation and GPU tuning
• Distributed systems at scale
• Owning systems end to end

This is our own read, written from the employer's posting and the company's Spotlight. The employer's posting is the source of truth for the role.

About Encord

A data platform that helps AI teams manage, label and clean the video, sensor and image data that trains robots, self-driving cars and other physical AI systems.