AI Trainer (Conversational AI/LLM Optimisation)

Remote contract | Outside IR35 | £350 per day | Initial 6-month term with planned extension to 12 months

Project snapshot
Help a major healthcare client launch a WhatsApp-based AI chatbot hosted on Microsoft Azure. You'll steer the continual learning loop that keeps the bot accurate, safe, and clinically relevant.

Key responsibilities
. Curate, label and augment real chat data to fine-tune a large-language-model (Azure OpenAI, Hugging Face, etc.)
. Design and run evaluation experiments (precision/recall, BLEU, CSAT, etc.), then turn findings into actionable model tweaks.
. Optimise prompts, retrieval-augmented-generation (RAG) chains and guard-rails to meet healthcare compliance standards (GDPR, ISO 27001, NHS DSPT).
. Define a repeatable training pipeline (MLflow or Azure ML) and performance dashboards for business stakeholders.
. Champion Responsible-AI best practice - bias checks, explainability, audit logs and drift monitoring.
. Work hand-in-hand with the chatbot dev team on release cadences and incident response.

Essential skills

Hands-on fine-tuning of LLMs and prompt-engineering in Python.

*

Azure ML, Cognitive Services, or comparable cloud ML stack.

*

Data labelling tooling (Label'studio, Prodigy, Scale, or similar).

*

Statistical testing, experiment design and curation of training corpora.

Desirable

*

Experience with healthcare vocabularies (SNOMED, FHIR) or NHS IG.

*

Familiarity with WhatsApp Business API