Data Scientist

ConnexAI; Manchester; On-site; GBP 70,000 - 110,000; Full time

Job summary

ConnexAI is seeking a Data Scientist (LLMs & Conversational AI) to design and evaluate production systems powered by large language models and real-time AI workflows. You will work with researchers and engineers across LLM behaviour, prompt optimisation, and evaluation to push the boundaries of conversational AI.

The role suits strong quantitative backgrounds in Mathematics, Physics, Statistics, Computer Science, or related fields, with a passion for rigorous problem solving in fast-paced,

Qualifications

• Degree/MSc/PhD in a quantitative field.
• Experience in data science, experimentation, statistical analysis, or ML systems.
• Curiosity about LLMs, NLP, or conversational AI.

Responsibilities

• Design datasets, experiments and evaluation methods for conversational AI solutions.
• Identify opportunities to improve quality, reliability and performance through research.
• Statistically investigate how changes to prompts, pipelines, models, or logic affect outcomes.
• Think about edge cases and test ways to handle them without harming global performance.
• Develop tests and metrics capturing user requirements and avoid proxies.
• Advance building blocks behind LLM-powered systems: classification, retrieval, scoring, and workflows.
• Collaborate with engineering/evaluation teams to translate product problems into AI problems.
• Stay up-to-date with research and apply ideas to production AI systems.

Skills

Python

Pandas

NumPy

Scikit-learn

Education

Quantitative degree (Math/Physics/Stats/CS)

Job description

Data Scientist (LLMs & Conversational AI)

About the role

We are building production conversational AI systems powered by large language models, speech AI, and real-time AI workflows.

We’re hiring two Data Scientists to join our LLM team, helping design, improve, and evaluate the systems that power our Conversational AI products in production. The focus is on defining, interrogating and solving complex modelling and evaluation problems, and building the data and experimentation foundations that will generate systematic advances in conversational AI.

You’ll work closely with researchers, engineers, and evaluation teams across areas such as LLM behaviour and reasoning, prompt optimisation, conversational quality, and model evaluation.

We’re particularly interested in people from strong quantitative backgrounds including Mathematics, Physics, Statistics, Computer Science, or related fields, who enjoy piecing together difficult problems, thinking rigorously, and working in environments driven by research and experimentation.

What you’ll do

• Design and use datasets, experiments and evaluation methods to gain detailed and intimate knowledge of our conversational AI solutions.
• Identify opportunities to improve quality, reliability, and performance through research and experimentation
• Statistically investigate how changes to prompts, pipelines, models, or system logic impact real-world outcomes
• Think creatively about edge cases that might expose flaws in a system, and devise and test different ideas about how to handle those cases without harming global performance
• Develop tests and metrics that capture the many nuances of user requirements, rather than uncritically relying on industry-standard proxies that might obscure underlying issues
• Advance the building blocks behind LLM-powered systems, including classification, ranking, retrieval, scoring, and conversational workflows
• Collaborate with engineering and evaluation teams to turn product problems into measurable AI problems
• Stay up to date with the latest research and critically apply new ideas to production AI systems, separating out strong, flexible approaches from fragile hype

What we’re looking for

• Degree, MSc, or PhD in Mathematics, Physics, Statistics, Computer Science, or a similarly quantitative field
• Strong Python skills with experience using tools such as pandas, NumPy, and scikit-learn
• Experience working with data science, experimentation, statistical analysis, or machine learning systems
• Curiosity around the finer details of LLMs, NLP, or conversational AI
• Ability to think critically about model behaviour and system performance rather than simply training models
• Sceptical mindset: you are uncomfortable with taking any assumption as given, and you examine new ideas with an open but critical mind
• Comfortable working in fast-moving, collaborative environments with ambiguous problems
• Work on real-world conversational AI systems used in production
• Solve difficult applied AI problems in a highly collaborative environment
• Join a research-driven team working across LLMs, speech AI, and evaluation
• Influence how AI systems are measured, improved, and deployed at scale