Overview

Foundation models can generate impressive images in seconds; enterprise teams still need outputs that match a company's visual language, follow production rules, and are safe to publish. exactly.ai gives creative teams a private, model-agnostic layer for knowledge, generation, and evaluation, turning brand rules and human feedback into reliable AI workflows.

Our customers include global e-commerce and technology platforms, major retailers, consumer brands, and enterprise software companies. We are named among the top 100 AI startups in the UK, and we are expanding our ML team to productise research that makes creative AI usable in production. We are looking for a Senior Applied AI / Multimodal AI Engineer who can turn ambiguous creative workflows into reliable AI systems.

You will work at the intersection of generative image models, LLM/VLM agents, evaluation systems, brand intelligence, and production engineering. The role combines fast prototyping, rigorous evaluation, reusable software design, and core product development: you will use real customer workflows and product usage to discover and validate capabilities, then turn what works into shared platform primitives, eval harnesses, agent workflows, and production-ready APIs.

You should be as interested in shaping the core product as in solving high-signal customer problems. You will build systems that understand brand guidelines, generate/edit images, preserve style and character identity, evaluate outputs, and learn from approvals, rejections, edits, and user experience without regressing on existing quality. A solution proven in one workflow should become a reusable pattern for other customers, models, and systems wherever it generalizes.

You'll report directly to the Head of AI, with broad ownership across the stack — from early prototypes to production workflows running for enterprise clients.

Compensation and work arrangement

• Compensation includes salary and equity; salary is determined based on the market benchmark and the candidate's years of experience.
• The team is distributed across Europe and remote work is fully supported; London-based candidates can also work from our office.
• CET is the team's main working timezone, so candidates should be comfortable working primarily on CET hours.
• The role offers 28 days of holiday in total, including bank holidays (using the bank holidays in your country if you are not based in England and Wales).

What you'll do:

• Build and evaluate AI workflows across image generation/editing, style transfer, visual QA, multimodal pipelines, and LLM/VLM-assisted agents, optimizing for quality, brand fidelity, consistency, safety, cost, latency, and regression risk.
• Turn successful prototypes and product insights into reusable production code, internal tools, eval suites, prompts, policies, and shared capabilities that can improve the core product and compound across customers, models, and workflows.
• Design feedback and evaluation systems that convert approvals, rejections, edits, production traces, and human review into measurable improvements while protecting existing performance.
• Collaborate with product, design, ML, and client-facing teams to shape roadmap decisions and translate subjective creative feedback into reproducible constraints, rubrics, and generation logic.
• Keep the team current on the AI landscape and help decide which new models, agentic workflows (including tool use, multimodal reasoning, planning, and orchestration), and evaluation methods are worth adopting.

Required Skills and Experience

• Strong Python engineering, with experience turning notebooks, scripts, and exploratory work into reusable, tested, production-ready AI workflows.
• Practical expertise with image-generation, image-editing, and computer-vision pipelines: image-to-image, style transfer, inpainting, character/style consistency, and multimodal reasoning around visual outputs.
• Ability to design evaluation systems for generative AI: golden datasets, visual comparison grids, LLM/VLM judges, human review rubrics, regression prompts, failure taxonomies, measurable quality gates, and feedback loops.
• Practical experience building LLM/VLM-assisted or agentic workflows: tool use, structured outputs, planning/orchestration, human-in-the-loop review, or automated QA.
• Strong spoken English, with the ability to discuss technical tradeoffs and subjective creative feedback clearly with colleagues and customers.

Bonus points for:

• Strong product, visual, and brand judgment: can reason about illustration systems, packaging, typography, layout, brand consistency, vague client goals, output failures, and creative feedback translated into technical constraints.
• Startup execution style: can move quickly from ambiguity to useful prototype while knowing when a workflow needs hardening before production use.

How to Apply

Please submit your CV or LinkedIn profile together with links to relevant work to artem@exactly.ai. This could include a GitHub repository, portfolio, technical article, open-source contribution or a short case study.

exactly.ai welcomes applicants from all backgrounds. We encourage candidates to apply even if they don't meet 100% of the listed requirements, and we will provide reasonable adjustments during the hiring process.