RL Systems Engineer – Hybrid Infra for Large-Scale AI

Roc Search; Greater London; Hybrid; GBP 70,000 - 110,000; Full time

Job summary

Roc Search is assisting a London-based AI startup in finding an Infrastructure Engineer to design and scale the deployment layer of a distributed platform. You will orchestrate complex workloads across on-premise hardware, ensuring reliable execution within memory and compute constraints while collaborating with research and engineering teams.

Key duties include building testing environments, scaling pipelines, and turning experimental concepts into production software, with a focus on

Qualifications

• Production-level Python development with clean testing and modular design.
• Experience with distributed computing and state synchronization.
• Memory and compute resource management across local tasks.
• Telemetry, logging and alerting in distributed or containerized environments.

Responsibilities

• Design runtime isolation and task scheduling for multiple local processes.
• Build robust reconciliation mechanisms for atomic updates across remote environments.
• Architect deployment flows with progressive rollout, validation, safety guardrails.
• Develop large-scale virtualized environments to emulate real networks for validation.
• Construct fault-tolerant distributed processing networks with automated recovery.
• Profile and optimize performance; tune resources and hardware acceleration.

Skills

Python

MLops

Pytorch

Grafana

Elastic Search

TensorRT

Distributed computing

Systems engineering

Resource management

On-premise hardware

Job description

Roc Search is assisting a London-based AI startup in finding an Infrastructure Engineer to design and scale the deployment layer of a distributed platform. You will orchestrate complex workloads across on-premise hardware, ensuring reliable execution within memory and compute constraints while collaborating with research and engineering teams.

Key duties include building testing environments, scaling pipelines, and turning experimental concepts into production software, with a focus on