Applied Scientist

i3; Greater London; Hybrid; GBP 120,000 - 180,000; Full time

Job summary

i3 in London is seeking a Lead Applied Scientist to drive Generative AI and agentic systems across complex problems. You will stay hands-on, build models in Python, and mentor data scientists and AI engineers while shaping technical strategy for real-world deployments.

You will design and evaluate LLM and agentic solutions, set experimentation standards, contribute to Responsible AI, and work closely with engineering leadership on architecture and productionisation in a hybrid London setup.

Qualifications

• Extensive hands-on expertise with LLMs and Generative AI.
• Strong experience with prompt engineering, RAG, fine-tuning and evaluation.
• Hands-on design and evaluation of agentic AI systems across multi-step workflows.
• Proficient with Databricks, MLflow and Spark-based data processing.
• Experience in Responsible AI practices and model risk.
• Ability to mentor other scientists and engineers.
• Evidence of taking AI solutions from prototype to production.
• Ability to explain complex AI concepts to stakeholders.
• Insurance/reinsurance data domain knowledge.
• Postgraduate quantitative degree.

Responsibilities

• Lead the scientific design and development of AI and machine learning solutions.
• Design sophisticated Generative AI and Agentic AI systems with multi-step workflows.
• Define evaluation methods, benchmarks, and acceptance thresholds.
• Apply rigorous mathematics and statistics to experimentation and validation.
• Own solution quality and production readiness.
• Solve challenging problems involving unstructured data and forecasting.
• Select appropriate approaches for each use case including LLMs and traditional ML.
• Build production-quality solutions and monitor for drift and reliability.
• Develop monitoring and observability for AI systems.
• Mentor Data Scientists and AI Engineers; collaborate with engineering leadership.
• Communicate methods and trade-offs clearly to senior stakeholders.
• Champion Responsible AI practices across projects.

Skills

LLMs & Generative AI

Prompt engineering

RAG

Agentic AI

ML/DL frameworks

Statistical grounding

Stakeholder comms

Education

Postgraduate degree in CS/Math/DS

Tools

Databricks

MLflow

Spark

pandas

NumPy

scikit-learn

Job description

Lead Applied Scientist – Generative AI & Agentic Systems

City, London/ Hybrid – 3 days per week in the office

We are looking for an exceptional Lead Applied Scientist to join the Data Science & AI function of a major international organisation investing heavily in Generative AI, Large Language Models and Agentic AI.

This is a senior, hands-on position for an accomplished Applied Scientist who combines deep expertise in modern AI with strong mathematical and statistical foundations.

You’ll act as a scientific authority for AI, setting the standards for how solutions are designed, evaluated, validated and monitored. You’ll work on some of the organisation’s most challenging AI problems while providing technical leadership and mentorship to Data Scientists and AI Engineers.

The focus is increasingly on LLMs and agentic systems, including RAG, tool use and multi-step agent workflows. However, this role is fundamentally about selecting the right scientific approach for each problem, drawing on machine learning, deep learning, statistics and optimisation where appropriate.

This is not a purely advisory or leadership position. You’ll remain highly hands-on, building models and AI applications, writing Python, designing experiments, reviewing code and solving technically challenging problems.

Your responsibilities will include:

• Leading the scientific design and development of AI and machine learning solutions, translating complex business problems into measurable technical objectives.
• Designing sophisticated Generative AI and Agentic AI systems, including prompting, RAG, tool use and multi-step agent workflows.
• Defining how LLM and agentic solutions are evaluated, including metrics, test sets, benchmarks, acceptance thresholds and evaluation frameworks.
• Applying mathematical and statistical rigour to experimentation, uncertainty, error analysis and solution validation.Acting as the technical escalation point for particularly complex or ambiguous AI problems.
• Holding scientific ownership for solution quality and ensuring approaches are robust before deployment.
• Solving challenging problems involving unstructured documents, expert workflow automation, forecasting, optimisation, portfolio analytics and claims analytics.
• Selecting the appropriate approach for each use case, whether that involves LLMs, agentic AI, deep learning, traditional machine learning or statistical methods.
• Building production-quality solutions directly, particularly for novel, technically challenging or higher-risk projects.
• Developing approaches to improve the accuracy, reliability and robustness of LLM and agent outputs.
• Defining monitoring approaches to identify model degradation and drift once solutions reach production.
• Leading Responsible AI practices, including bias and fairness testing, explainability and validation of model and agent behaviour.
• Setting standards for experimentation, evaluation, coding and technical documentation.
• Reviewing code, experiments and technical outputs produced across the wider team.
• Mentoring and developing Data Scientists and AI Engineers through pairing, technical reviews and knowledge sharing.
• Working closely with engineering leadership on architecture, deployment and productionisation.
• Communicating technical methods, results, limitations and trade-offs clearly to senior business stakeholders and governance teams.

You should bring:

• Deep practical expertise with Large Language Models and Generative AI.
• Strong experience with prompt engineering, Retrieval Augmented Generation (RAG), fine-tuning and LLM evaluation.
• Hands-on experience designing and evaluating agentic AI systems, including agents that use tools and operate across multi-step workflows.
• Experience with modern AI/ML and data science libraries and frameworks, alongside technologies such as pandas, NumPy and scikit-learn.
• Strong mathematical and statistical foundations, including probability, statistics, linear algebra and optimisation.
• The ability to reason rigorously about uncertainty, error, model behaviour and statistical significance.
• Strong machine learning fundamentals, including model validation and experimental design.
• Experience creating evaluation and validation frameworks for AI systems, particularly LLM, RAG and agent outputs.
• A proven track record of taking AI solutions from prototype through to production.
• Practical experience with Databricks, MLflow and Spark-based data processing.
• Knowledge of Responsible AI practices, including bias, fairness, explainability and model risk.
• Experience setting technical or scientific standards and developing other scientists and engineers.
• The communication skills to explain highly complex AI concepts to both technical and non-technical senior stakeholders.
• Agent frameworks and orchestration technologies for building multi-step, tool-using AI systems.
• Advanced approaches to LLM and agent evaluation.
• AI observability and production monitoring.
• Experience within insurance or reinsurance, particularly involving underwriting, claims or actuarial data.
• A postgraduate qualification in Computer Science, Statistics, Mathematics, Data Science or another quantitative/computational discipline.