TL;DR
AI Engineer (AI): Designing and refining production-grade LLM agents and algorithms for legal litigation, focusing on NLP fundamentals, structured outputs and cost-efficiency. Focus on AI strategy, development of evaluation frameworks and pipelines.
Location: Excited to work in our office in London. We work from the WeWork in Aldwych, and we spend 4 days a week in person.
Salary: £70K – £100K
Company
Wexler is building the best AI system for litigation on the planet, working with some of the world’s largest law firms to solve their most complex cases.
What you will do
- Design, build and refine production-grade LLM agents and proprietary algorithms for expert litigators
- Lead development of robust evaluation frameworks and pipelines, establishing metrics and benchmarking models on large datasets to steer R&D
- Drive advanced prompt-engineering practice to maximise model efficacy
- Apply core NLP fundamentals (tokenisation, embeddings, model architecture) while keeping outputs structured, performant and cost-efficient
- Own clean, modular Python back-end code for data-intensive systems, balancing real-world constraints such as token limits, cost and latency
Requirements
- Expert-level Python with a track record of shipping production LLM systems
- Hands-on experience designing LLM agents and RAG pipelines
- Proficiency with AI evaluation frameworks and end-to-end evaluation pipelines
- Solid grounding in NLP fundamentals (tokenisation, embeddings, bias handling)
- Ability to juggle token limits, cost and latency while delivering structured outputs
Culture & Benefits
- Competitive salary and meaningful early-stage equity (£70-90K)
- Huge autonomy and and ownership. You will be designing and realising AI systems for some of the most expert users in AI.
- Budget for learning and professional growth
- Bi-annual team retreats
Будьте осторожны: если работодатель просит войти в их систему, используя iCloud/Google, прислать код/пароль, запустить код/ПО, не делайте этого - это мошенники. Обязательно жмите "Пожаловаться" или пишите в поддержку. Подробнее в гайде →