Senior AI/ML Engineer, London

Description

At vector8 we design, develop and deliver production grade AI Solutions, leveraging our deep understanding and specialism in Financial Services and Retail sectors.

Our solutions are varied depending on the problem we are solving, ranging from agentic systems to more traditional AI-powered solutions based on LLMs, classical ML or speech models.

Every solution meets the same impeccable standard of performance, robustness and long-term maintainability.

This is a hands-on engineering role where you will

• Design and build agents and AI workflows, including tool integration, orchestration, guardrails and human validation steps.
• Integrate LLMs, retrieval and classical ML or speech models with clients' existing systems and data.
• Make these solutions reliable in production through systematic evaluation, tracing, fallback strategies, and cost and latency control.
• Apply software engineering best practices: testing, CI/CD, modular design and documentation.
• Address the requirements of regulated industries, such as security, compliance, access control and integration with legacy systems.
• Own solutions end to end, from scoping with the client to production monitoring and continuous improvement.
• Work closely with vector8's engineers and project managers, as well as with client architects and business teams.

You will help build AI foundations that our clients can reuse and extend as they scale from individual use cases to company-wide capabilities.

The role is primarily based in London, with occasional travel to client sites and collaboration with teams across Europe.

Requirements

• 1. Agentic systems and AI workflows
• Design and build LLM agents that use tools and interact with enterprise systems and data
• Orchestrate multi-step and multi-agent workflows, with human validation and hand-off where needed
• Define clear boundaries for each agent through typed tool contracts, structured outputs, guardrails and access policies
• Connect agents to other systems through open protocols such as MCP and A2A, and through well-designed APIs
• Choose the most appropriate architecture for each problem, whether an agent, a deterministic workflow or a classical ML solution
• Evaluate AI solutions systematically against datasets to generate meaningful performance metrics.
• 2. Software Engineering for AI
• Write clean, modular, and well-documented code in Python (Fast API, Pydantic, asyncio).

• Apply best practices in

• Testing (unit, integration, end-to-end)
• CI/CD (Git Hub Actions, Git Lab CI, Argo CD)
• Observability (logging, monitoring, tracing).
• Ensure security and compliance (data protection, access controls, encryption).
• Integrate models and code into CI/CD pipelines for seamless deployment.
• 3. Enterprise AI & MLOps
• Deploy and monitor AI solution in production

• Design and implement MLOps pipelines for

• Model training, fine-tuning, and evaluation
• Model versioning and lineage tracking
• A/B testing and canary deployments.
• Ensure scalability and reliability (auto-scaling, fault tolerance, disaster recovery).
• Collaborate with data engineers to build data pipelines (batch, streaming, real-time).
• 4. Collaboration & Technical Leadership
• Work closely with product owners, Dev Ops, and quality assurance in an agile, cross-functional team.
• Mentor junior engineers and promote best practices in AI/ML and software engineering.
• Translate product requirements into technical solutions and architectural decisions
• Document architectures, decisions, and best practices for internal and client-facing use.
• Develop relationships with internal and external stakeholders, including clients and partners.
• 5. End-to-End AI-powered Solutions Development (less frequent but can occur)
• Design, implement, and deploy distributed, high-volume, high-performance, low-latency machine learning solutions, with a focus on Gen AI models, and especially LLM integrations and API-driven architectures.

• Take ownership of your models throughout their entire lifecycle

• Data exploration and cleaning to build reproducible, versioned datasets
• State-of-the-art research to identify the best architectures for the problem (e. g., transformers, RAG, fine-tuning).
• Implementation, training, and optimization in reproducible environments
• Deployment, monitoring, and maintenance in production.
• Optimize models for performance, latency, and cost efficiency, especially in LLM serving and inference.
• 6. Innovation & Continuous Improvement
• Stay ahead of the latest AI and ML architectures (transformers, Mixture of Experts, sparse attention).
• Experiment with cutting-edge techniques (quantization, distillation, speculative decoding).
• Evaluate and benchmark open-source and proprietary models (Llama, Mistral, Mixtral, GPT-4, Claude).
• Bring your own ideas through vector8’s ideation process.
• Contribute to vector8’s AI accelerators (reusable components for common industry problems).
• Embrace a strategic and continuous improvement mentality to drive innovation.