Towards AI, Inc.
Senior AI Engineer / Forward Deployed Engineer
Requirements
Candidates need expert daily experience with agentic coding tools and at least two shipped AI systems beyond simple chat, with the ability to explain their architecture, evaluation, failure modes, costs, and rollout. Strong Python and production engineering experience are required, typically including 5+ years in software engineering and 2+ years building production LLM applications, along with sound judgment on reliability, security, privacy, latency, and cost.
Job Description
London · full-time · reports to the co-founders. You will own the technical delivery of custom AI systems end to end, from architecture through build, evaluation, hardening and production rollout, for portfolio companies and finance firms that cannot afford to get AI wrong. Most weeks are heads-down engineering. On our largest engagements you are forward deployed: attached to the client for part of the week, surfacing the next thing to automate with the strategist.
- Own the codebase and the engineering plan of an engagement: from a spec or a feature, decide how it is built, and own both the architecture of the code and the architecture of the AI system.
- Build production LLM applications: agentic systems for investment research and due diligence, retrieval-augmented copilots, document, image and video analysis, workflow automation for portfolio operations.
- Make evals a deliverable (golden sets, regression tests, human review loops, release gates, production monitoring) and set the guardrails for privacy, prompt injection and hallucination risk.
- Deploy and run on GCP, Azure or AWS with Docker, CI/CD and observability; hand the repo, tests and documentation to the client's team, or keep extending the system.
- Mentor the associates and engineers paired with you, and ship reusable firm assets: reference architectures, eval harnesses, agent skills, MCP servers.
Daily, expert use of Claude Code, Codex, Cursor or similar agentic tools is non-negotiable; engineers who work this way outpace the rest by a wide margin, and that gap is the bar. We expect intelligent supervision: you give the agent the right context, constrain the task, run the tests, catch weak assumptions and take control back when needed. You never ship what you don't understand, and you build the tooling (skills, MCP servers, eval harnesses) that makes the team faster.
A builder and a field engineer: you take an ambiguous technical problem, find the constraint that matters and design a system people actually use, and you know when an agent is the right shape, when deterministic code is better and when retrieval is the bottleneck.
Required
- Daily, expert use of Claude Code, Codex, Cursor or similar agentic tools, with a setup you have tuned and the habit of inspecting what they produce; you will show us.
- At least two shipped AI systems beyond simple chat, and the ability to walk us through one in concrete terms: architecture, model choices, retrieval design, evals, failure modes, cost, rollout.
- Strong Python and production experience with APIs, integrations and data-heavy systems; real agent depth (tool calling, structured outputs, orchestration, state, retries, safe execution boundaries).
- Production judgment on latency, cost, reliability, privacy and security; Docker, Git, CI/CD, cloud deployment; clear technical communication, to engineers and to a client's CFO.
- Typically 5+ years of software engineering, including 2+ building LLM applications in production.
Useful: PE or portfolio-company work; multimodal or vision work; deep RAG; solutions or customer engineering; TypeScript, React, Postgres, LangGraph or Claude Skills.
The senior level of our engineering ladder. Engineers who want client exposure are attached to clients in the forward-deployed mode; the next step leads the technical side of a portfolio-wide programme, or takes a platform and security remit across engagements.
- Location: our London office, fully in person, with flexibility when you need it; some travel to clients in the UK and abroad.
- Right to work: you need the right to work in the UK before you start; for exceptional candidates we can sponsor a visa.
- Pay: a market-rate base calibrated to the level, plus a potential performance bonus.
- Tools: team plans on Claude Code and Codex; Granola, Notion, Slack and Linear; the Towards AI curriculum.
- We hire for talent, curiosity and character, and want a team where people of every background can do their best work.
Two to three rounds depending on the role, 45 minutes each; we aim for about three weeks from first conversation to offer: a conversation with two co-founders about what you have built and how you build with AI; a take-home case capped at five hours, delivered as a repo, then a debrief where you present it (no live coding tests); for strategists and enablement leads, a live case or workshop in our London office where possible; then a call with one of our co-CEOs to make the offer, subject to two references and right to work.
Towards AI Deployment is the build side of Towards AI Inc., the AI deployment and education firm founded in 2019 (500,000+ practitioners trained, a 200,000+ newsletter, the O'Reilly book Building LLMs for Production). We go into private equity firms, portfolio companies and finance firms and ship AI systems they can actually run: not pilots that quietly die, but tools that hold up in production and change how the business works. We are 20+ people, with experience from Palantir, BCG, UBS, J.P. Morgan, SAP and Mila.
Apply through linkedin or the form on our website at towardsai.com. A link to what you have built and a few lines on your most useful AI work tell us more than a long CV.
Skills
About Towards AI, Inc.
We turn AI from a tool people experiment with into a capability teams rely on. Towards AI runs two connected ventures. Towards AI Learning has trained 500,000+ AI practitioners since 2019 through our courses, book, publication, and newsletters (200,000+ subscribers), and converts software developers into AI engineers and forward-deployed engineers. Towards AI Deployment puts them to work inside private equity firms, portfolio companies and funds, building custom AI agents and automation in pods with deployment strategists who joined us from Palantir, BCG, UBS, J.P. Morgan, and Aurelius Growth Capital. Production lessons flow back into what we teach, and our best students become the engineers who build. We believe AI initiatives fail when organisations lack the skills, education, and deployment capability to realise value, so we train your team while we build.