Applied AI engineer in beauty and wellness

Fresha; Greater London; Sur place; GBP 90 000 - 130 000; Plein temps

Résumé du poste

Fresha — ведущая AI‑платформа для индустрии красоты и велнес, для которой требуется опытный backend‑инженер с фокусом на AI. Ваша роль — владение AI‑фичами от идеи до выпуска, проектирование сервисов и интеграций вокруг модели, обеспечение надежности и низкой задержки в реальном времени.

Требуется 5+ лет опыта в продакшене, работа с LLM/API и агент‑фреймворками, навыки проектирования контекста и инструментов, понимание распределённых систем и оптимизации затрат.

Qualifications

• 5+ лет разработки и эксплуатации backend‑сервисов в продакшене.
• Опыт внедрения решений на основе LLM API или агент‑фреймворка с реальными пользователями.
• Знание дизайна контекста и инструментов для агентов; умение оценивать некон deterministических систем.
• Понимание распределённых систем, расходов и задержек; прагматичный подход к решениям.

Responsabilités

• Владение AI‑фичами от идеи до выпуска: проектирование, сборка, оценка, развёртывание и поддержка в реальном общении с клиентами.
• Построение системы вокруг модели: сервисы, интеграции и инструменты для агентов.
• Разработка оценочных наборов, града́торов и production‑проверок для измерения влияния изменений.
• Сотрудничество с Product, Design, Support и другими командами на протяжении цикла выпуска.

Connaissances

LLM integration

Distributed systems

Cost & latency optimization

Evaluations

Tool design & retrieval

Backend development

Kafka

Outils

LLM APIs

Agent framework

Monitoring & tracing

Description du poste

Описание:

Fresha is the leading AI-powered operating system for the global beauty and wellness industry, giving businesses everything they need to run smarter, grow faster and build stronger client relationships.

Задачи:

Own AI features end to end: Design it, build it, evaluate it, ship it, and look after it once real customers are talking to it. Build the system around the model: The services, tools and integrations an agent needs to do real work: checking availability, making bookings, taking payments, escalating to a human. Correct when it retries, safe when it's wrong. Prove it works: Build the evals, graders and production checks that tell us whether a change made the product better. Treat them as part of the product, with their own accuracy you measure and improve. Engineer the context: Decide what the model knows, what tools it has and how they're described. Most quality gains come from here, not from the prompt. Work in real time: Voice and chat run on tight latency budgets. You'll balance model choice, cost, latency and quality, and keep all four in view. Learn from real conversations: Read transcripts, instrument runs end to end (traces, cost, latency, outcomes), and turn what you find into the next change. Work with the rest of the team: Product, Design, Support and other engineering teams, from the first conversation through to release. Have a say in how we build: Our architecture, our AI practices and our engineering practices aren't fixed. We expect senior engineers to push on all three.

Требования:

Experience: 5+ years building and running backend services in production. Built with LLMs, not just used them: You've shipped a product on top of LLM APIs or an agent framework that real users depend on, and you've had to make it reliable. You know where models fail and how to design around it. Evals: You've written evaluations for non-deterministic systems: test sets, graders, LLM-as-judge, and you know when to trust each. You'd rather measure than guess. Context and tool design: You understand that what an agent knows and what tools it can call matter more than the prompt. Retrieval, tool design and MCP are familiar territory. Correctness: You think about what happens when a request arrives twice, a downstream call times out halfway through, or an agent takes an action it shouldn't. You design for it rather than finding out later. Distributed systems: You've built and run services at scale, and you know how they fail. Cost and latency: You know that the right model for a task depends on the task, and you keep spend and response time in view. Product sense: You care whether customers get what they need, not just whether the code is clean. You're happy reading transcripts and talking to the people who use what you build. AI in your workflow: You use AI coding tools day to day and have a clear view of where they help and where they don't. You've built your own workflows, and you review what comes back before it ships. Languages: Strong in at least one backend language. We don't expect Elixir on day one, but we do expect you to pick things up fast. Pragmatic: You'd rather use what exists than build something new, and you can tell the difference between a demo and a product people rely on. At Fresha, we value passion and potential as much as specific skills. Будет плюсом: Event-driven design with Kafka or equivalent is a plus

Условия:

AI that customers actually use: Your work talks to real people every day, at the front desk of thousands of businesses. You'll know quickly whether it helps. Scale: Around 1 million bookings a day across 120 countries, and the payments that go with them. The actions our AI takes have real consequences. Speed: We release daily, 500+ deploys a day, with not much standing between your branch and your users. Ownership: A small team with the tooling, data and permissions to run what it builds, and a say in where the product goes next.