Applied AI / ML Engineer

Penta; Greater London; Hybrid; GBP 55,000 - 90,000; Full time

Benefits offered by this job

Competitive salary

Paid time off

Social events

Learning environment

Location-based benefits

Job summary

Penta is seeking Applied AI / ML Engineers early in their careers who are excited by AI-native software development. This hands‑on role focuses on designing, building, testing and deploying practical AI workflows using Python, LLMs, AWS and Claude Code.

You will build and operate real systems in AWS via Claude Code, drive infrastructure and contribute to data science delivery with Python, SQL and APIs. The team values practical builders who ship working software and learn quickly, with coaching

Qualifications

• Hands-on experience building with at least one modern AI-native tool or platform (e.g., Claude Code, OpenWebUI, LiteLLM).
• Strong Python coding ability to build, read, test and debug real software.
• Solid prompting and context‑engineering skills and familiarity with LLM workflows.
• Comfort with APIs, JSON, structured outputs and everyday developer tooling.
• Basic to intermediate SQL.
• Git version control.
• Bias towards shipping working software, not just analysis or notebooks.

Responsibilities

• Build AI-native tools and workflows using Claude Code and frontier AI tools.
• Create practical LLM workflows, agents, skills and tools to support teams and client delivery.
• Implement reusable AI capabilities using OpenWebUI, LiteLLM and MCP tools.
• Work across prompting, context engineering, structured outputs, evaluation, provider routing and human‑in‑the‑loop workflows.
• Move useful applications from prototype to production and ensure code is maintainable and safe.

Skills

Python

Claude Code

OpenWebUI

LiteLLM

AWS

SQL

Git

Prompting

Context Engineering

MLOps

Docker

Tools

Claude Code

OpenWebUI

LiteLLM

MCP tools

Job description

What’s in it for me?

• Competitive salary and compensation structure
• Generous paid time off and holiday schedule
• Frequent firm‑wide social events and activities
• Excellent environment for learning and growth
• Further benefits, depending on location

About the Role

We are looking for Applied AI / ML Engineers, early in their careers, who are excited by the current generation of AI-native software development. This is a hands‑on builder role: you will help design, build, test and deploy practical AI workflows and data science capabilities using Python, LLMs, AWS and Claude Code. The most important attribute is that you already use modern AI tools to build faster and better. Claude Code is our primary way of building, so we want people who are fluent and fast with it, with the judgement to know when to check and use its output and when to change it. You should be equally keen to build and operate real systems in AWS, driving that infrastructure through Claude Code itself. You may come from a data science, machine learning, software engineering or technical STEM background. Traditional data science skills such as experimentation, model evaluation, data analysis and pipeline development are useful and welcome, but they are not the current centre of gravity for this role. The core of the job is building: AI-native tools, automated workflows, and the infrastructure that turns ideas into deployed capabilities. What you’ll do: Build AI-native tools and workflows

• Use Claude Code and other frontier AI tools to accelerate development, prototyping, debugging and documentation.
• Build practical LLM workflows, agents, skills and tools that support advisory teams, client delivery and internal operations.
• Implement reusable AI capabilities using OpenWebUI, LiteLLM and MCP tools.
• Work across prompting, context engineering, structured outputs, evaluation, provider routing and human‑in‑the‑loop workflows.
• Move useful applications and automations quickly from prototype to production.
• Review and improve AI-generated code, not just generate it: read it, test it, and make it maintainable and safe to ship.

Build and command AWS

• Stand up, deploy and operate the AWS services your tools run on, working with EC2, ECS, S3, Lambda, IAM and CloudWatch.
• Increasingly drive and operate AWS directly through Claude Code, infrastructure-as-code and agentic tooling.
• Support automated deployment, monitoring, logging, cost control and observability.
• Apply DevOps and MLOps practices pragmatically, working with DevOps and engineering colleagues to keep systems reliable, maintainable and easy to operate.

Support applied AI and data science delivery

• Use Python, SQL and APIs to build applied AI and analytics capabilities.
• Help process and enrich large volumes of text and unstructured content.
• Contribute to workflows involving summarisation, classification, topic identification, sentiment, contextualisation, embeddings, vector search and RAG.
• Help evaluate AI outputs for quality, reliability, cost and usability, and document approaches clearly so others can understand, trust and reuse them.

Learn, improve and contribute

• Take ownership of defined tasks, prototypes and production improvements.
• Take direction well and level up quickly; you will be coached hard and expected to become faster and more independent over time.
• Join hackathons, experiments and rapid delivery cycles, and stay curious about new AI tools, models and agentic patterns.
• Share what you learn and help raise Penta’s practical AI capability.

The Team The Data Science team sits within Penta’s Technology function, alongside Development, Engineering / DevOps & Platform Security, IT and the PMO, and works hand in hand with Product and the advisory teams. We are rebuilding and expanding the team around a practical, AI-native roadmap. The focus is not academic research or traditional data science in isolation; it is building useful, reliable AI-enabled tools and workflows that improve how Penta serves clients and how our internal teams work. You will work under the VP / Head of Data Science, contributing hands‑on to applied AI systems, LLM workflows, agentic tools, reusable skills, internal automation and production‑ready capabilities. The environment is collaborative, pragmatic and delivery‑focused. We move quickly, share ideas openly, and value people who can turn ambiguity into working software. You will build a lot, ship often, and be coached closely as you grow.

Our approach to AI is practical

We are forward‑thinking and ambitious in practically applying AI across client‑facing work, corporate teams, technology and data science. We do not train our own foundation models; we apply, orchestrate and evaluate the best available models from major providers, engineering them into robust workflows that solve real business problems.

We validate and document our methodologies through white papers and technical explainers, setting out the academic and analytical foundations behind our products, and the evidence that supports their use. This helps build trust in our tools, why they work and shows how they can be applied.

We build on open‑source AI software, including LiteLLM and OpenWebUI, to maximise the internal value of AI and build durable institutional capability. Reliability, cost control, observability, instruction‑following, usability and adoption matter most.

The team designs reusable workflows, skills, models and MCP tools that support Penta’s move towards an AI‑native advisory operating system. Keeping Penta at the forefront of practical AI in PR, communications and strategic consultancy is part of the role.

Success means you are helping Penta ship useful AI and data science capabilities into real workflows. Within the first few months, you should be contributing to working prototypes, internal tools, AI workflows, AWS deployments and production improvements. Over time, you should become increasingly independent in building reliable, maintainable AI‑native systems that improve client delivery and internal productivity.

Our Values

• Empathy and collaboration
• Pushing our ideas
• Facing adversity
• Ownership and leadership

About You

Technical skills We do not expect candidates at this level to have deep experience in every area. We are looking for strong potential, practical ability, and clear evidence that you already build with modern AI tools. Hands‑on experience with at least one AI‑native building tool or platform is mandatory, and we will want to see examples of what you have made with it.

Essential

• Hands‑on experience building with at least one modern AI‑native tool or platform, for example Claude Code, Codex, OpenWebUI, LiteLLM or Cowork / OpenCowork (or similar), with concrete examples of what you have built with it.
• Strong Python coding ability: enough to build, read, test and debug real software.
• Strong prompting and context‑engineering skills, and familiarity with using LLMs in real workflows.
• Comfort with APIs, JSON, structured outputs and everyday developer tooling.
• Basic to intermediate SQL.
• Git version control.
• Strong debugging, problem‑solving and self‑learning skills.
• A bias towards shipping working software, not just analysis or notebooks.

Highly desirable

• Breadth across AI-native tools, MCP tools and agentic frameworks.
• Experience building small apps, internal tools, automations or backend services, and with Docker or containerised development.
• Experience with embeddings, vector databases, RAG or semantic search, and with evaluating LLM outputs for quality, accuracy, cost and reliability.
• Experience processing large volumes of text or unstructured content.

Nice to have

• Any hands‑on AWS experience (EC2, ECS, S3, Lambda, CloudWatch, IAM, infrastructure‑as‑code or CI/CD) is a plus. You will also develop this on the job, increasingly by driving AWS through Claude Code.
• Foundational machine learning knowledge, and classic NLP such as classification, clustering or model selection.
• Experience with hypothesis‑led data science.
• A technical degree in Computer Science, Software Engineering, Data Science, Machine Learning, Maths, Physics or a related field. Welcome, but not required if the building track record is there.

Key competencies

• A practical builder who enjoys turning ideas into working tools.
• AI‑native in how they work, using AI tools as a normal part of development.
• Curious, fast‑learning and comfortable with ambiguity.
• Technically rigorous without being academic for its own sake.
• Sound in their judgement about AI output: healthily sceptical, and in the habit of testing and verifying rather than trusting blindly.
• Comfortable asking questions and learning from senior colleagues.
• Able to communicate technical ideas clearly to technical and non‑technical people.
• Focused on usefulness, reliability, cost and maintainability.
• Excited to help build an AI‑native advisory operating system.