Freelance Senior AI Engineer
At a glance
Senior AI Engineering role at Projects by IF. London · contract.
Pay not stated AI Engineer salary
aimlroles summary, based on the employer's posting.
What you'll do
- Build prototypes that let agents use services through APIs, MCP servers and agent-to-agent protocols
- Create complete agent systems, connecting orchestration, tools, memory and state
- Set up safeguards, evaluations and monitoring to check agents behave safely and correctly
- Test agent-led user journeys alongside designers and researchers, using real users
- Explain feasibility, risks and standards, then turn prototypes into reusable patterns, documentation and code
What you bring
- You can design and build agent-facing service prototypes using APIs, MCP servers and agent-to-agent protocols
- You can build agent systems end to end, including orchestration, tool connections, memory and state
- You can develop safeguards and methods to evaluate and monitor whether AI systems act safely and correctly
Who this fits
This suits a hands-on senior engineer who can build quickly, assess the implications of technical choices and explain trade-offs to non-technical people. You’ll work on a six-month contract, five days a week, with at least two days in the London office. The work is exploratory and involves collaboration with designers, researchers, civil servants and government technical teams.
From the employer
Role Details:
Level: Lead
Start Date: December
Duration: 6 months
Day Rate: Dependent on experience
Weekly hours: 5 days/ week
Location: Hybrid, 2 days a week in London
We're looking for a Senior AI Engineer to help us reimagine a function of government as part of an agentic ecosystem. You'll join a small multidisciplinary team of designers, researchers and engineers, working closely with civil servants and the department's own technical teams.
The work is exploratory. We will learn by building and iterating, leaning on emerging best practices and patterns. We need someone who can build working prototypes quickly, while thinking deeply about the implications of technical decisions.
Key skills and experience
Must haves:
Design and build prototypes that expose services to agents through APIs, MCP servers and agent to agent protocols
Design and build agentic systems end to end, including orchestration, tool integrations, and memory and state
Build guardrails, evaluations and monitoring to check agents act safely and correctly
Work with designers and researchers to test agent-led journeys with real users
Advise on technical feasibility, risks and standards, in plain language
Turn prototypes into patterns, documentation and code that other government teams can reuse
Nice to have:
Substantial software engineering experience, with recent hands-on work building LLM-based applications or agents
Strong knowledge of tool use, agent frameworks and orchestration, and protocols such as MCP, and API design
Practical experience with guardrails, evaluation, testing and monitoring of AI systems
Sound judgement on security, privacy and safety risks, including prompt injection and misuse
Comfort working in fast, uncertain early-stage work, and explaining technical trade-offs to non-technical people
Location
You will need to be able to work from our offices (London) for a minimum of two days a week.
1. Application
Please attach your CV, plus a portfolio or any work you would like to also share.
2. Initial screening call
A short conversation to cover the basics — your timezone, working languages, availability, and a chance for you to ask early questions about the role and how we work.
3. Technical interview (~60 minutes)
A deeper conversation about your technical background and experience. We’ll talk through how you’ve approached real problems, the decisions you’ve made, and how you’d apply that thinking to the work at IF.
4. Ways of working interview (~45–60 minutes)
A conversation about how you collaborate, communicate, and work with others — including how you operate in an advisory capacity with clients and teams.
We’ll aim to keep you updated at every stage and let you know the outcome promptly after each step.