Forward Deployed AI Engineer

Location: London, United Kingdom.

Contract: Freelance, with the possibility of future full-time employment.

About the job:

AI is forcing engineers to evolve like never before. The role is increasingly defined by deciding what a system should do and which specs and evals to write, then handing execution to agents and carefully checking what comes back. The best engineers we know still understand every layer underneath and are eager to spend more time thinking about how and why the solutions they build get used.

That’s who we want to work with at Calibr, a startup founded by Peter Wood and Marlene Leiss, and part of The Apprentify Group, one of the UK’s leading AI and tech training providers. We build tailored AI solutions for companies going through AI transformation, then help their teams build the skills to run, extend and own those systems themselves. As a Forward Deployed AI Engineer, you’ll work directly with clients to scope the problem, build the solution and leave them more capable than when you arrived.

Who we are:

Our mission at Calibr is to help companies build in-house AI capability. Our approach is simple: help businesses identify where AI can create the most value, prove it in a focused part of the business and build the capability to navigate AI adoption long term.

We are industry-agnostic and work with the UK’s leading SMEs, scale-ups, corporates, NGOs and public sector organisations. Our group’s clients include Jet2, JD Group, Canon, BBC, B&Q, McLaren and Manchester Airport, to name a few.

The Apprentify Group is a collective of six specialist businesses, including Netcom Training, Flourish, Ioda, The Juice Academy and Apprentify. We are united by a shared commitment to providing learning opportunities to people at every stage of their career, and in 2025 we were named the UK’s 30th-fastest-growing private company by The Sunday Times.

Who we are looking for:

We are looking for a technically competent orchestrator (i.e. someone who can explain the decisions behind their code and develop it in ways that respect wider architectures) to help build transformation projects for clients.

Your goals will be to turn a scoped business problem into well-bounded technical compositions, build tested and evaluated guardrails, and create feedback loops that involve humans, resulting in AI-assisted delivery that is efficient yet explainable and trustworthy.

What you will do

• Decompose ambiguous problems into smaller components and map out their dependencies and operational risk before production.

• Determine the right set of tools (harnesses, models, custom tools and MCP servers).

• Code the underlying plumbing and necessary automations to enable the design of the system while embedding it in a pre-existing architecture.

• Build observability channels to monitor the health of the system and allow for user-accessible fixes.

• Bound your builds with schemas, evaluation sets and permissions for human review to enable transparency.

• Explain architecture decisions, their trade-offs and risks in terms non-technical stakeholders can act on.

What you bring

We understand that this is a new and emerging role, so we focus far more on the signals we receive from you as a candidate than on the job title on your CV. You are, however, likely to be a match if you have demonstrable competence in the following areas:

• Building stress-tested cloud systems that have made it to production and had measurable impact (3+ years).

• Shipping AI harnesses involving RAG, document intelligence and custom tools, with a focus on evaluation pipelines, model routing, and agent/tool orchestration and permissions.

• Working with Python (including FastAPI), specifically LangChain (and LangGraph where necessary), SQL and scalable (vector) databases.

• Enforcing rigorous evaluation to cover edge cases and determine specific mitigations for prompt injection and tool abuse.

• Distinguishing a model safety issue from an application security issue.

Travel: This role may require being on-site with clients.

How we work:

As part of The Apprentify Group, we have the backing, clie