Full Stack AI Engineer

Xist4 IT Limited.; City Of London; Teletrabalho; GBP 85 000 - 105 000; Tempo integral

Resumo da oferta

Xist4 IT Limited is hiring a Full Stack AI Engineer for a fully remote role across the UK. You will own features from frontend to AI workflow using Next.js, Python, and Node.js in a production product.

You will design AI-driven workflows with LLMs, memory, and tools, ensure reliability, latency management, and real-time interactions, and work in a small distributed team alongside ML, backend and product engineers.

Qualificações

• Experience shipping full-stack features from idea to production.
• Hands-on frontend and backend coding in real projects.
• Shipped LLM, RAG or agent-based applications.
• Experience API and system design in production.
• Multistep AI workflows with tools, memory, or orchestration.
• Proven ability to diagnose production failures and latency.
• Right to work in the UK.

Responsabilidades

• Build frontend, backend, and AI integration features end-to-end.
• Design workflows that plan, use tools, and recover across steps.
• Integrate LLMs, memory, and external tools into product behaviour.
• Ensure reliability with fallbacks and recovery for failures.
• Ship features and monitor real-world performance.

Conhecimentos

Full stack development

Frontend and Backend

LLM integration

API design

AI workflows

Production reliability

Ownership after release

UK work authorization

Ferramentas

Next.js

Python

Node.js

PyTorch

Kubernetes

Docker

SQL

NoSQL

Descrição da oferta de emprego

Full Stack AI Engineer | Next.js, Python, LLMs | Fully remote, UK

£85,000 to £105,000. Permanent.
London. Fully remote across the UK.

This is full stack in the literal sense. You will build the interface, backend and AI workflow, then stay with it when an agent stalls, a tool fails or the model returns something you did not plan for.

Our client is an early-stage AI product company building applications that get on with everyday tasks before you ask. A prototype exists, launch is ahead, and the company is funded without relying on an upcoming round.

You'll work in a small, distributed team with ML, backend and product engineers. You own a feature from the first screen to the last service call, and the AI sits inside the product, not behind a separate research hand-off.

The hard part is reliability. Multi-step workflows cross models, memory, tools and services. Streaming, partial results, latency and recovery have to be designed into the feature from the start. Adding them after launch doesn't work.

The job

Product. Build features across frontend, backend and AI integrations. The current stack includes Next.js, Python and Node.js, with SQL and NoSQL storage behind the product.
Agents. Design workflows that plan, use tools, retain context and recover across multiple steps. You'll integrate LLMs, memory and external tools into product behaviour that can be observed and improved.
Reliability. Build fallbacks and recovery around model and tool failures. You'll work on streaming and real-time AI interactions where latency, partial results and user feedback need deliberate handling.
Production. Ship the feature, watch how it behaves and iterate from real failure modes. Docker and Kubernetes are part of the stack, alongside OpenAI, Anthropic and open-source models.

What you'll bring

Essential:

• Full stack features you have taken from idea to production.
• Frontend and backend code you have both written yourself.
• LLM, RAG or agent-based applications you have shipped.
• API and system design used in a production product.
• Multi-step AI workflows involving tools, memory or orchestration.
• Production failures or latency problems you have traced across the whole stack.
• Work where you owned the feature after release, not only delivery.
• Existing right to work in the UK.

Useful:

Next.js, Python, Node.js, PyTorch, Kubernetes, Docker, SQL, NoSQL.

Who this suits

You are probably a Full Stack Engineer, Product Engineer or AI Engineer who likes owning the whole feature. You are comfortable moving from a user interaction to an API, an agent workflow and back again.

It won't suit you if you want the model to be somebody else's problem. Here the agent that stalls halfway through a task is your feature, and so is the fix.

Let's talk

Actively looking or just curious, we welcome applicants from every background and will support reasonable adjustments.