Who we are
Blackstone& are a technical consultancy who specialise in digital transformation and complex technical deliveries. 15 years ago that meant Cloud adoption. 10 years ago it was DevOps. 5 years ago, SRE and platform engineering. We still cover all of that, but you guessed it, a lot of our work right now is AI adoption. And we mean actual AI adoption.. in production... in Enterprises. Not just a vibe coded app on a local machine.
The pace and impact of each wave keeps accelerating, but we think successful adoption still comes down to two things: people and constraints. What does transformation actually look like inside the limits of an organisation, and how do we bring the people along for it?
Interested in helping us answer those questions for our customers?
The role
This is a hands-on contract embedded in live customer delivery. Youll build and ship AI systems in complex, high-stakes environments, not run proof-of-concepts that quietly die before they reach production. From the architecture of RAG pipelines to the evaluation harnesses that prove they work, the work is yours to own.
Youll work closely with delivery managers, solution architects and client stakeholders, turning real business problems into AI that does the job. Youll bring engineering discipline to a space that often lacks it, and help shape how we deliver AI at scale.
What were looking for
* Solid back-end development. You write clean, production-grade code that other people can maintain, test and build on.
* Real experience building and running evaluation harnesses to measure LLM performance, output quality and regression.
* A deep, working understanding of large language models: how they behave, where they break, and how to engineer around it.
* Proven RAG experience, including chunking, embedding choices and retrieval tuning.
* Enterprise client exposure. Youre comfortable in complex stakeholder environments with governance, compliance and security in the mix.
* Hands-on with Azure AI Foundry or equivalent Azure AI services, covering deployment, monitoring and integration.
* Strong context engineering, designing prompts and context windows that are robust and repeatable.
* A systematic approach to prompting. You test, iterate and document rather than guess.
Nice to have
* Production AI agents: tool use, memory, and orchestration with the likes of LangChain, Semantic Kernel or AutoGen.
* Familiarity with agentic design patterns, including multi-step reasoning, planning loops and safe agent behaviour in regulated settings.
* Fine-tuning or model adaptation for domain-specific use cases.
The details
* £competitive
* Outside IR35
* Hybrid, a mix of remote and on-site with the client as the engagement requires. Negotiable.
* 6 months to start, with extensions expected based on performance and how the project develops
* Start date to be agreed, though wed like to move quickly for the right person
If you take engineering seriously, stay curious about AI without losing your scepticism, and like the idea of your work being visible in a small team, wed like to hear from you.