Stealth Startup
Head of Applied Legal – AI Product (Fractional)
Get notified of more jobs like this · No spam, ever
How your CV stacks up
Upload your CV to see how well it fits this job role
?%
Early-stage AI startup | London (Hybrid) or UK remote | Fractional contract | Around 2 days per week, open to discussion | Day rate depending on experience | Start October 2026
THE SHORT VERSION
We are hiring a senior person to own the legal logic inside an AI product: someone who has led, not supported, the work of turning legal and regulatory obligations into plain-English rules, controls and tests that a system applies. You will set the framework, make the calls, and tell the founders what good looks like. We are not looking for someone to hand tasks to; we are looking for the person who decides what the tasks are. You may be a lawyer, a privacy engineer, a compliance engineer, a policy engineer or a regulatory-content lead. We care about what you have led and built, not your title.
Fractional, around two days a week as a starting point, though we are open on the number of days subject to scope and your availability. Paid at a day rate, agreed depending on experience. UK hours with some Monday–Wednesday overlap. Initial eight weeks with scope to continue and to grow as we raise and scale.
THE PROBLEM YOU WILL WORK ON
Organisations want the productivity of autonomous AI agents but are afraid to deploy them, because they cannot show what those agents are permitted to do or that the limits hold. In practice an agent's authority is a large set of technical permissions (thousands of actions across cloud and workplace systems), and the law it engages is written in a different language entirely.
Our product translates the technical layer into terms a lawyer can reason about (a class method becomes "read", a queue write becomes "send externally"), links configurations of agent permissions to the specific obligations they engage, and finds configurations that reduce legal exposure. Software gets a customer most of the way; a human expert makes the final judgement on what remains. You own the design of the bridge in the middle: the framework, the first rule library, the review model and the standard everyone after you follows.
A concrete example: one agent may read customer documents; another may post to a shared message board and send email externally. Each permission is reasonable on its own. Together they create a pathway that engages data-protection and confidentiality obligations. You define the plain-English rule that catches that pathway, the facts the rule needs, the exceptions, the green / amber / human-review outcome, and what a passing check does and does not prove.
WHAT YOU WILL OWN IN THE FIRST EIGHT WEEKS
You lead this workstream. The CTO builds what you specify; the founders and the expert legal panel are your sounding board, not your supervisors.
- Design the obligation-to-control framework: how legal and regulatory requirements are identified, written as plain-English rules, mapped to system actions and permissions, and kept current as law and facts change. It must extend from the first use case (UK financial services) to other sectors and jurisdictions.
- Build the first rule library for one bounded workflow, working directly with the CTO and with our expert legal panel: obligations in, testable rules and control requirements out, with sources, assumptions and limits documented. Distinguish legal requirements from customer policies and commercial choices, and keep them separate in the model.
- Design the green / amber / red decision logic and the human review around it: referral triggers, the information a reviewer needs, and how interpretations, source changes and review outcomes are recorded so they can be reused.
- Define what a passing check establishes and what it leaves unproven, so the product's outputs and our customer-facing claims never go beyond the evidence.
- Set the standard for the expert reviewers who follow you and for how their judgement is captured and reused.
Reasons to use Rodeo
Start with a chat, not a search bar
Grad scheme, placement, apprenticeship? Not sure what you want yet — that's fine. Your agent talks it through with you and turns "I have no idea" into a shortlist.
It searches the market for you
Every day your agent scans the market matching roles against what actually matters to you, not just keywords on a CV.
Only hits
No noise. No "maybe this fits." Just roles with a clear explanation of why they're right — and where to focus when applying.
THE INITIAL ASSIGNMENT
We will agree one bounded workflow and a manageable set of obligations before you start. By the end of the initial eight weeks we expect:
- A shared description of the agent activities in that workflow, the legal questions they raise and the facts needed to assess them.
- Documented decision criteria with sources, assumptions, exceptions and unresolved questions, in a structure that can be extended to the next workflow and jurisdiction.
- Representative test cases, and an assessment of whether the prototype reflects the intended interpretation.
- Clear boundaries for human review, and a plain statement of what the assessment can and cannot establish.
Success means the reasoning is understandable, testable and useful to the technical team, with uncertainty made explicit, and we know what is feasible and what needs a different approach.
WHO YOU MIGHT BE
We have seen the work we need done in at least six places. If you recognise yourself in any of them, apply.
- The privacy or compliance engineer: you have built data-flow controls, purpose-limitation checks or automated policy enforcement inside a product or platform, working from what the lawyers told you the obligation was.
- The legal engineer or legal knowledge engineer: you have encoded legal logic into a rules-based or AI product at a legaltech or regtech company, or in a law firm's legal-engineering team.
- The policy engineer or applied-policy specialist: you have turned usage policies and legal constraints into enforceable rules, classifiers or decision trees at an AI company or platform.
- The regulatory-content or regulatory-intelligence lead: you have decomposed regulation (FCA, PRA, GDPR, EU AI Act) into structured requirements and controls for a regtech product or a bank's control framework.
- The GRC or controls architect: you have mapped legal and regulatory obligations to control libraries (ISO 27001, SOC 2, NIST) and built the tests that evidence them.
- The rules-as-code practitioner: you have worked on encoding legislation or policy as executable rules in government, public-sector digital or a rules-engine vendor.
Lawyers with any of the above are especially welcome; lawyers without it, and engineers without any exposure to legal obligations, are probably not the fit.
WHAT WE NEED FROM YOU
- Typically 10+ years in legal, privacy, compliance or regulatory-technology work, with evidence that you personally led the design of a rule set, control framework, policy-enforcement system or product feature that turned obligations into something a system applied, and were accountable for how it performed. This is the one thing we will test hard.
- The seniority to guide two founders through an unfamiliar domain: you set the agenda for your workstream, say clearly what cannot be done, and do not need handholding.
- Fluency in at least one relevant body of law in practice: data protection and privacy, financial-services regulation, or AI-specific regulation. You do not need to be the expert in all of them; our expert panel covers depth. You need to know how obligations are read and where the edges are.
- Technical literacy: you can read a permissions table or a data-flow diagram, understand roles, scopes and access controls, and argue about edge cases with engineers. You do not need to write production code.
- Comfort with bounded, imperfect systems: you can make uncertainty explicit and design the caveats and referral rules that make an automated judgement honest.
- Availability from October, in the UK time zone, including part of Monday to Wednesday to work with the CTO. We have two days a week in mind, but the commitment is open to discussion depending on scope and what you can offer.
Get help with your application
Your very own career expert that helps elevate your application to the next level.
Strongly preferred: a legal qualification (England and Wales or equivalent) or significant time inside a legal or compliance function. Useful: cloud IAM experience (AWS, Azure, GCP), and experience with LLM tooling.
ABOUT THE COMPANY
We are a UK-based, founder-led AI startup, pre-launch. The founding team combines deep experience recruiting scarce senior expertise with a technical lead who has spent a decade putting AI models into production in regulated domains. Around the founders is a panel of senior lawyers who provide the legal judgement the product relies on; you will work with them, not replace them. Pre-revenue, working prototype, first paid design partners in UK financial services being lined up now. The ambition is a system of record for consequential AI across sectors and jurisdictions. Product detail is shared at first conversation under NDA.
WHAT WE OFFER
- A paid, fractional engagement at a day rate agreed depending on experience, with the number of days shaped around scope and availability, reviewed after the initial period, with scope for deeper involvement and upside as the company reaches its next commercial and funding milestones.
- Ownership of a mapping problem no one has templated yet, working directly with the CTO, in a product whose value depends on your judgement. This is a founding-level seat, not a support role.
- A small, candid team that would rather hear "that cannot be expressed as a rule" than a confident guess.
HOW TO APPLY
Apply through the LinkedIn job listing only. Send your CV or profile, two short sanitised examples of obligations you turned into rules or controls, your availability and preferred fee basis. No cover letter.
We welcome independent practitioners and people in existing roles who have capacity for an agreed engagement alongside their current commitments, subject to any relevant permissions or conflicts.
25-minute first call with the CEO, then 45 minutes with the CEO and CTO working through one real mapping example together. Any substantive practical assignment is agreed and paid. Product detail is shared at the first call under a short mutual NDA.
“It took my CV and asked me questions relevant to understanding what kind of jobs to suggest for me. Suggestions were almost perfect. Jobs were exactly what I’ve been looking for.”
Jessica, London
Location