Build useful AI without separating capability from responsibility.

Depending on the role, you may:

• —Understand complex organisational problems
• —Design and build AI products
• —Develop reliable production systems
• —Evaluate models and complete systems
• —Integrate AI into real workflows
• —Build knowledge and retrieval infrastructure
• —Design human review and safeguards
• —Conduct technical and applied research
• —Work directly with clients and domain experts
• —Turn recurring problems into products
• —Create open-source or public-interest technology
• —Improve how the company itself operates

The work may move between research, product development, engineering, implementation, and institutional design. We are interested in people who can go deep in their field while understanding how their work fits into the larger system.

We want to work with people who:

• —Care whether a system works beyond the demonstration
• —Begin with the problem, not the technology
• —Can distinguish evidence from confidence
• —Communicate limitations clearly
• —Take reliability, security and privacy seriously
• —Work with people outside their own discipline
• —Raise disagreements directly
• —Share knowledge and document important work
• —Accept responsibility for outcomes
• —Will recommend that no AI system be built
• —Want to participate in the value they create
• —Pursue commercial success without treating growth as the purpose

You do not need to agree with every idea held by the founders. You do need to be able to work honestly within the company's published commitments.

We care more about the quality of your judgement and work than about polished self-presentation. Evidence may include products or systems you have built, research, open-source contributions, technical writing, design work, client or organisational outcomes, difficult problems you have helped resolve, systems you have maintained, failures you examined honestly, and knowledge you helped others develop.

We value both original creation and the less visible work that makes systems dependable, including testing, maintenance, documentation, monitoring, security, and support.

Mission alignment does not compensate for an inability to perform the work. Technical strength does not exempt someone from responsibility for its consequences. Depending on the role, we may assess:

• —Technical depth
• —Product judgement
• —Quality of reasoning
• —Understanding unfamiliar domains
• —Communication
• —Operational discipline
• —Identifying failure modes
• —Collaboration
• —Ownership of outcomes
• —Revising a position when evidence changes

We are not looking for people who merely speak confidently about AI. We are looking for people who can build, evaluate, and operate it responsibly.

How we work

We begin by understanding the problem. We ask whether AI is necessary. We build the smallest system capable of testing the important assumptions. We evaluate it under conditions resembling real use. We make responsibility visible before deployment. We document important decisions, limitations, and failures. We tell clients when they do not need what they are asking for. We refuse work that depends on deception, manipulation, exploitative surveillance, compulsive engagement, or the unnecessary removal of human judgement.

Initial conversation

We discuss your experience, the work, the company, and whether there appears to be a meaningful fit. This is also an opportunity for you to examine whether the company's structure and commitments are acceptable to you.

Evidence of work

We review work that helps us understand how you think and perform. This may include past projects, research, writing, code, designs, systems, or a detailed discussion of a decision you made.

Role assessment

You may complete a technical, product, research, design, or operational assessment relevant to the role. We avoid irrelevant puzzles and unnecessarily long unpaid assignments. Where an assessment requires substantial original work that could create value for the company, it should be compensated or replaced by another method.

Working discussion

You may work through a realistic problem with people you would collaborate with. We are interested in how you clarify the problem, identify assumptions, consider alternatives, communicate uncertainty, respond to disagreement, and balance speed, quality, and responsibility.

Mutual decision

Before an offer, both sides should understand the role, responsibilities, reporting relationship, compensation, participation, location and working expectations, company stage and uncertainty, applicable policies, and potential work across K4M2 AI and K4M2A Foundation. The process should help both sides make a more informed decision, not pressure a candidate into accepting quickly.

Early-stage companies contain uncertainty. Roles may change. Priorities may move. Products may fail. Revenue may be uneven. Some systems and policies will still be under development.

We should tell candidates clearly what has been legally established, what is currently practiced, what remains an intention, what the company can afford, what risks the role carries, what authority the role actually has, which decisions remain with founders, executives, the board, or the Foundation, and what may change as the company develops.

We do not want candidates to join an imagined future company while working inside a materially different present one.

We may work with people across locations and time zones. Each role should state clearly whether it is remote, hybrid, or location-specific, the required working-hour overlap, travel expectations, the employment or contracting structure, applicable geographic restrictions, and communication and availability expectations.

Remote work should not imply permanent availability. People should understand when immediate response is required and when asynchronous work is expected.

We want people with different backgrounds, disciplines, geographies, languages, and forms of experience to be able to contribute meaningfully. We do not want inclusion to become a substitute for clear standards. People are expected to perform their work competently, communicate honestly, treat others with respect, protect confidential information, disclose relevant conflicts of interest, raise material risks, accept direct feedback, avoid retaliation, take responsibility for commitments, and follow the company's safety and governance requirements.

A respectful culture must still be capable of addressing poor performance, misconduct, and repeated failure to meet responsibilities.

Employees, candidates, contractors, and collaborators should have a way to raise concerns about hiring practices, working conditions, compensation, discrimination, harassment, retaliation, conflicts of interest, product safety, misleading claims, privacy or security, and work that conflicts with the company's published boundaries.

People raising concerns in good faith should not be punished for doing so.

Open roles