Applied AI Engineer

**Seamflow** Applied AI Engineer; £100K–£150K base per year + meaningful equity (~0.1%–1%); London (On-site)

**About Company** Seamflow builds AI-native software for the Testing, Inspection, and Certification (TIC) industry — a ~$300bn market that sits between manufacturers and regulators, verifying that products, processes, and systems meet safety, quality, and regulatory standards. It's a large, legacy industry built during the Industrial Revolution: still highly manual, resistant to change, and only now feeling real pressure to adopt AI. Seamflow builds a suite of tools that remove the manual bottlenecks in these workflows, and is one of the first companies doing it properly — rather than an off-the-shelf product, they build tools and customise them in granular detail to each customer's workflows.

Seamflow raised a 14x oversubscribed $4.5M seed round earlier this year from Initialized Capital and Northzone — the funds behind the founders of Coinbase, Reddit, Klarna, and Spotify — plus angels. They're already deployed inside most of the largest TIC players, seeing rapid revenue growth (demand outstripping supply), and plan to raise a Series A before the end of the year. The market is Euro-centric (hence the London HQ) but expanding into the US and China.

The team is ~11 people, scaling toward ~20. Engineering is world-class: a founding engineer from Fuse Energy (a ~$5bn company, where he was a first hire), people from X (working directly with Elon Musk), Amazon, Yandex, and JetBrains, plus AI researchers from MIPT, UCL, and Johns Hopkins. Commercially, the team includes the youngest-ever associate partner at McKinsey, managers from Bain, and Forbes 30u30 alumni. CTO Yusuf leads the technical vision; lead engineer Dimitri (ex-Fuse, first hire) sits beneath him, and the structure is deliberately flat — engineers sit together daily, argue out the best outcome, and every voice is heard.

**About the Role** One of two Applied AI Engineer seats, at the core of Seamflow's product. For someone who thinks in user workflows and model behaviour — not benchmarks or isolated prompts — and wants to decide what AI should do, build it end to end, and put it in users' hands. You'll own LLM-powered features from prompt to evals to production: structured extraction from messy real-world documents, making models reliable in a regulated domain, and agents/orchestration for complex TIC workflows. Any exceptional engineer here sits across the whole team and owns their vision.

**Responsibilities** - Build LLM-powered features end to end, from prompt to eval to deployment. - Ship an AI feature to production in the first week. - Read hundreds of real model outputs, debug edge cases, and talk directly to customers. - Build evals that catch real regressions rather than vanity metrics. - Decide which model to use and how to make it reliable; diagnose why models fail. - Build multi-step agents, tool use, and orchestration for complex domain workflows. - Extract structured data from messy PDFs, scans, forms, and tables.

**Required Skills** A. Exceptional engineering pedigree (non-negotiable) — standout work at a top company, leading AI lab, elite trading firm, or fast-scaling startup. A famous logo isn't required, but it helps. B. Strong Python and modern LLM tooling — OpenAI, Anthropic, and Azure OpenAI SDKs; structured outputs; function calling. C. Has built production AI systems that go beyond a wrapper around a chat API, with hands-on frontier-model experience. D. Comfort with evals, prompt engineering, and model failure modes. E. Works across the stack — TypeScript/React frontend, Python backend. F. Reasons about complex domain workflows, not just prompt-and-response, with strong product judgement.

**Bonus Skills** - Published papers or research with frontier models (nice-to-have, not a must). - Founder or ex-founder; fellowships such as EF, YC, or Antler. - World-class CS/engineering education. - AI products shipped in production at an early-stage or Series A/B company.

**Logistical Info** - Location: London, in-person at the Shoreditch HQ. Default in-office but flexible ("if you've got to be at home a certain day, that's fine") — outcome-focused. Open to relocation from anywhere. - Compensation: £100K–£150K base (set by experience, not role) + meaningful equity (~0.1%–1%). - Benefits/Other: Full visa sponsorship and relocation incl. flights; all meals in-office (breakfast, lunch, dinner); free gym membership.

**Ideal Background** - Strongest signal — leading AI labs / scale-ups: Mistral, Cohere, Perplexity, ElevenLabs, Isomorphic Labs, DeepMind, Aleph Alpha, Poolside, LangChain, DeepL, Synthesia, Inflection, Hugging Face, Helsing, Black Forest Labs, Anthropic, Unlikely AI, METR, Modal Labs, InstaDeep. - Also strong — big tech: JetBrains, AWS, Yandex, ByteDance, Datadog, Uber, Microsoft AI, Google, BCG X. - Also strong — trading firms (AI/infra calibre): Jane Street, Optiver, Citadel, DRW, HRT, Jump, Two Sigma, XTX, IMC, G-Research. - Also strong — high-ownership startups: Harvey, Happy Robot, Physical Intelligence, Cursor, Ramp, Anduril, Linear, Bending Spoons. - Can lean research-driven, founding-engineer, or lab-based — what matters is calibre and a real AI project that impresses a Seamflow engineer. Big tech straight out is fine if paired with founder mindset.

**Green Flags** - Founder mindset: high agency, high ownership, low ego. - Builds end to end at production grade and ships fast. - Thrives in unstructured, ambiguous problems — "builds order out of chaos." - Intensity — willing to work long hours, including nights/weekends when needed.

**Red Flags** - No founder mindset; only comfortable in a structured/corporate environment. - Multiple short stints (e.g. several ~6-month roles). - Contract roles rather than full-time. - Misleading or irrelevant job titles (e.g. "data trainer" at a big tech firm rather than a genuine SWE role). - Shiny logos without real depth or impact.

**Interview Process** Fast by design; fully remote — candidates may come on-site if they wish. 1. CV submission → reviewed almost immediately by Andrea Olivo (Founder's Associate). 2. Screening call (~30 min) with Andrea — motivations, early-stage startup appetite, willingness to work hard, past experience. 3. Technical conversation (~40 min) with a founding engineer — not an exam; an in-depth technical discussion. 4. Take-home task (~2 hours). 5. Final "vibe check" with both founders to sign off.