Software Engineer III

Trusted Connectivity for AI-Powered Dominance. Everfox delivers trusted connectivity to protect the world’s most critical environments and safeguard the sensitive data powering decision advantage. Built for mission-critical operations, Everfox protects what matters most by securing how data moves, how users access it, and how threats are neutralized across every domain. We enable mission speed and secure collaboration across networks, domains, and allies, while ensuring the data powering AI and advanced analytics remains trusted and protected. We don’t just defend systems; we deliver decision dominance. Our Purpose. Protect what matters most by enabling organizations to operate security in the most complex and high-risk digital environments. Our Vision . To be the most trusted authority in high-assurance cybersecurity, enabling secure collaboration across every domain. Our Belief. Security should never limit mission success – it should enable it. Our Promise. We deliver absolute confidence in environments where failure is not an option. Everfox is hiring a Software Engineer to join our team in London, UK. This role suits a strong developer who enjoys landing in unfamiliar code, working out how it fits together, and making it better. You will work on a rolling series of projects across the business, each typically lasting one to six months, adding AI-driven features to our products and using AI tooling to improve how we write, test and ship software. Every project is different. You might be building an AI-assisted capability into a customer-facing product, standing up evaluation and guardrails around a model integration, introducing agentic workflows into a team's development process, or using AI tooling to lift test coverage and cut manual effort in a legacy codebase. The languages, frameworks and platforms will change with each project. What counts is the quality of your reasoning and problem-solving, your ability to get productive in a new codebase quickly, and how you take on a hard challenge, not the exact languages on your CV. The codebases you work in are often owned by other internal teams. So a large part of the job is working with those expert engineers: understanding their constraints, agreeing an approach, writing code that fits their standards, leaving behind something they can maintain and build on after you move to your next project, and ultimately generating sharable artifacts that will benefit other projects. Done well, this role spreads good practice across the whole engineering organization, and your work is visible quickly across multiple products and teams. If you like variety, learning fast, and being the person who brings a new capability into a team rather than doing it in isolation, this role is for you. Location : This is a hybrid role in London, UK. Key Responsibilities: Engineering Design, implement and maintain features and components across a range of internal codebases, in whichever languages and frameworks each project requires. Add AI-driven capability to our products: model and API integration, prompt and context design, retrieval, tooling and agent orchestration, plus the plumbing around it such as data handling, caching, error handling, cost and latency management, and graceful behaviour when a model is unavailable or wrong. Build evaluation and testing around AI features so that behaviour is measurable rather than anecdotal, and quality does not regress as models, prompts and dependencies change. Use AI tooling and agentic workflows in your own day-to-day development to improve speed and quality, and build the reusable pieces that let others do the same: Skills, slash commands, subagents, hooks, MCP integrations and per-codebase agent guidance, tailored to each team's stack and conventions. Apply the same tooling to the codebases you work in: raising test coverage, generating and maintaining tests, improving documentation, and reducing repetitive manual work. Prototype quickly to de-risk the harder or less well understood problems, and be willing to conclude that an approach is not worth pursuing. Get up to speed in unfamiliar code fast: read it, map the important parts, ask good questions, and start contributing without needing everything explained. Working with other teams Work as a guest in codebases owned by other teams: agree scope and approach up front, follow their conventions, and get changes reviewed and merged rather than left on a branch. Work with Product Managers, QA and the owning engineering teams to turn a broad ask into well-scoped technical work with a clear definition of done. Hand over cleanly at the end of each project. Leave tests, documentation, agent tooling and enough context that the owning team can support and extend what you built, and stay available for a sensible tail of questions. Share what you learn. Each project should make the next team's job easier, whether through reusable components and tooling, patterns, internal write-ups or informal coaching on AI tooling. Keep your progress, risks and blockers visible to your line manager and to the teams you are working with. Quality, security and improvement Take responsibility for the quality of what you ship: automated tests, code review, sensible logging, and thinking through failure modes rather than assuming the happy path. Apply appropriate care to security and data handling when introducing AI components, including what data is sent where, what a model is trusted to do, and how untrusted input and model output are handled. We build security products, and our own engineering is expected to reflect that. Contribute to CI/CD, automation and developer tooling improvements where they unblock the work in front of you. Help us build a shared view of what good looks like for AI in our products and in our development process, and feed back honestly on what works and what does not. Key Knowledge and Skills: A degree in Computer Science, Engineering, or a related technical or scientific subject, or equivalent practical experience. Solid commercial software development experience, with a track record of shipping working software that other people rely on. Genuine language flexibility: strong in at least one language, and comfortable becoming productive in others as projects demand. We care about your ability to learn a codebase, not your CV keywords. A good grasp of core computer science concepts: data structures and algorithms, concurrency, networking, APIs and interfaces, testing, and the basics of software security. The ability to reason about a system you did not build: read unfamiliar code, form a mental model, identify where a change belongs, and understand its knock-on effects. Practical experience using modern AI development tooling, or clear evidence that you have taught yourself new tooling and techniques quickly and applied them to real work. Strong communication and collaboration skills. Much of your impact depends on other teams trusting your work and adopting it. Self-motivated, flexible and adaptable, and comfortable being thrown in at the deep end on an unfamiliar project. Nice to have: Hands-on experience with agentic coding tools such as Claude Code, Codex or similar, used for real work rather than casual experimentation. Experience extending those tools rather than just using them out of the box: authoring reusable agent extensions (Claude Code Skills, slash commands, subagents, hooks, MCP servers) and the project-level guidance files (CLAUDE.md and equivalents) that make an agent effective in a specific codebase. A good chunk of this role is likely to involve building that kind of reusable tooling so each team gets the benefit after you have moved on. Experience integrating LLMs or other models into production software: APIs, SDKs, RAG, tool and function calling, agent frameworks, or local and self-hosted inference. Experience evaluating AI systems: building test sets and benchmarks, measuring output quality, or automatin