Data Scientist

About the role

We help leading brands such as Uber, Amazon and HelloFresh put customers at the centre of everything they do. We are looking for a Data Scientist to help build and ship the next generation of our Customer Experience Intelligence platform.

Key responsibilities

• Train, evaluate and iterate on ML models for customer‑feedback tasks, including custom fine‑tuning pipelines.
• Build and maintain LLM‑powered features such as retrieval pipelines, reranking systems and insight generation.
• Develop evaluation frameworks, create test sets and define metrics for classification, extraction and generative tasks.
• Work on semantic search and embedding‑based retrieval, extending beyond standard approaches.
• Write clean, well‑tested code and collaborate with engineering on model integration and monitoring.

Required profile

• Strong knowledge of transformer architectures applied to NLP.
• Proficiency with PyTorch, including training loops and fine‑tuning workflows.
• Experience handling real‑world text data for classification, extraction, embeddings or search at scale.
• Familiarity with instruction fine‑tuning or model serving concepts.
• Solid grounding in classical machine learning, statistics and the ability to choose simpler methods when appropriate.

Required skills

• Transformer architectures
• PyTorch
• LoRA (parameter‑efficient fine‑tuning)
• Text classification
• Information extraction
• Embedding‑based search
• Instruction fine‑tuning
• Model serving
• Classical machine learning
• Statistics

What we offer

• Flexible remote or hybrid work setup with a work‑from‑home allowance.
• 25 days holiday plus local bank holidays and a day off for your birthday.
• Annual learning and development budget that grows over time.
• Equity options to share in the company’s success.
• Monthly health and wellbeing budget, optional private healthcare and life assurance.
• Enhanced family, fertility and neonatal leave where applicable.
• Dog‑friendly London office with classes, events and a rooftop terrace.