AI Architect

AutoThink Group; Rochdale; Hybrid; GBP 120,000 - 180,000; Full time

Job summary

AutoThink Group seeks an AI Architect to own the technical blueprint across retrieval, evaluation, safety, routing, and cost governance for all products. You will set standards the group builds on and drive architectural decisions at scale.

You'll build a group-wide reference architecture for LLM, RAG and agent workflows, implement a model-routing layer, and develop evaluation harnesses to gate every production release while ensuring privacy and GDPR alignment.

Qualifications

• 8+ years engineering experience with 3+ shipping production LLM/ML systems.
• Deep knowledge of transformer models, RAG, vector search and agent frameworks.
• Experience designing systems at scale on AWS, GCP or Azure.
• Strong opinions on evaluation, safety and cost control with receipts.
• Excellent written and verbal communication; able to pitch to board and staff engineer.

Responsibilities

• Own the technical blueprint for AutoThink products across retrieval, evaluation, safety, routing, cost governance and orchestration.
• Develop a group-wide reference architecture for LLM, RAG and agent workflows.
• Create model-routing and cost governance layer to select providers per request.
• Build evaluation harnesses and offline benchmarks gating production releases.
• Define guardrails, PII handling and audit trails aligned with UK GDPR guidance.

Skills

LLM/ML systems

RAG & vector search

Agent frameworks

Cloud platforms (AWS/GCP/Azure)

Evaluation & safety

Cost governance

Effective communication

Tools

AWS

GCP

Azure

Vector databases

Model routing

Job description

As AI Architect you own the technical blueprint that underpins every AutoThink product from retrieval, evaluation and safety layers, through to model routing, cost governance and agentic orchestration. You set the standards the whole group builds on.

What you'll build

• A group-wide reference architecture for LLM, RAG and agent workflows deployed across 12+ live products.
• Model-routing and cost-governance layer that decides which provider serves each request.
• Evaluation harnesses and offline benchmarks that gate every production release.
• Guardrails, PII handling and audit trails aligned with UK GDPR and ICO guidance.

Requirements

• 8+ years engineering experience, 3+ shipping production LLM/ML systems.
• Deep knowledge of transformer models, RAG, vector search and agent frameworks.
• Track record designing systems at scale on AWS, GCP or Azure.
• Strong opinions on evaluation, safety and cost control with the receipts.
• Excellent written and verbal communication; you can convince a board and a staff engineer in the same afternoon.