Contract:3 months (extendable)

Location:Remote (UK-based candidates preferred)

IR35:Outside IR35​

We are seeking an experienced AI/NLP specialistto join on a 3-month extendable contract. The successful candidate will manage a small, capable team tasked with applying AI and natural language processing to perform Quality Assurance. We want someone who can bring practical experience and a track record of delivering AI/NLP projects in production. You’ll know what works, what doesn’t, and how to avoid common pitfalls. This is an opportunity to make a significant impact by helping an existing team move faster and smarter with applied AI/NLP in a real-world business context.

​

Responsibilities:

• Shape the approach to prompt design, evaluation, and optimisation using LLMs.

• Define and implement logic to summarise call transcripts and validate against QA checklists.

• Build an evaluation framework to test and measure model performance.

• Assess trade-offs between platforms and models (e.g. Snowflake Cortex vs AWS Bedrock) for cost, performance, and suitability.

• Work with existing internal experts to accelerate delivery and inject proven experience into the project.

Primary Skills

• Strong background in Python, SQL, dbt, Airflow.

• Hands-on experience with LLM prompt engineering (framing prompts, evaluating outputs, managing prompts with LangChain).

• Familiarity with Snowflake Cortex and AWS Bedrock.

• Ability to compare model performance/costs, including fine-tuning and optimising smaller models.

• Experience building logic for transcript summarisation and automated QA validation.

• Design and implementation of evaluation frameworks for AI/NLP solutions.

Nice to Have (Engineering Skills)

• Integrating LLM APIs to process text at scale via Snowflake Cortex or AWS Bedrock.

• Building APIs on AWS to connect systems (e.g. Slack, Excel, CRM platforms).

• Knowledge of data security and governance practices when working with PII data.
  

If this sounds like an opportunity you are interested in, apply now for an immediate review