AI Data Scientist

**AI Data Scientist** We're working with a fast\-growing, VC\-backed healthtech startup on a mission to build a platform that blends human insight with advanced AI for personalised, continuous care. As part of this growth, they’re looking for an AI Data Scientist to work within a distributed team, on a hybrid model. **Your Impact** As one of their earliest AI hires, you'll shape the backend foundation powering secure data flows, AI\-driven insights, and provider tools. Expect hands\-on ownership in a collaborative, mission\-focused team tackling real\-world health challenges. **Key Responsibilities** * Design and execute experiments using large datasets, including LLM\-generated or LLM\-augmented data * Analyze and validate the reliability, consistency, and bias of LLM outputs across healthcare use cases * Fine\-tune and evaluate LLMs (e.g., OpenAI, Claude, Llama) while managing risks like overfitting and hallucination * Build scalable pipelines to preprocess, structure, and extract insights from unstructured or semi\-structured data * Work closely with clinicians and product teams to align AI insights with real\-world healthcare needs * Develop metrics and evaluation strategies for model performance, safety, and explainability * Investigate and mitigate risks related to synthetic data and model\-induced artifacts * Help shape how healthcare AI tech can be safe, fast, and deeply human **What You'll Bring** * 5\+ years of experience in data science or machine learning, including work with LLMs or large generative models * Deep understanding of LLM internals\-tokenization, attention mechanisms, fine\-tuning, prompt engineering, embeddings * Strong Python skills, including libraries like PyTorch, HuggingFace Transformers, LangChain, or similar * Hands\-on experience fine\-tuning models or building applications with LLM\-generated data * Strong statistical and experimental design skills, especially around model evaluation and failure analysis * Familiarity with cloud\-based ML pipelines (e.g., AWS/GCP/Azure), versioned datasets, and reproducible experimentation * Experience navigating data quality issues\-bias, hallucination, inconsistency ***For further details and immediate consideration, please get in touch.***