Director Applied AI for Marketing and Commerce

About this role

<p><strong>Who We Are</strong></p> <p>Artefact is a leading global consulting company dedicated to accelerating the adoption of data and AI to positively impact people and organizations. We build systems that run in production, in our clients' own cloud, and keep running after we leave. 2,000+ people across 27 countries and 1,000 clients, including some of the largest consumer brands in the world. Not a strategy house that discovered AI eighteen months ago, and not a prototype shop that hands you something you will rebuild from scratch.</p> <p>Three ingredients make that possible, and very few firms have all three: the data, real and messy and at enterprise scale; an applied AI team that can put a model into a governed production environment; and the transformation muscle to make people actually use it.&nbsp;</p> <p>Agentic AI lands on marketing and commerce before it lands anywhere else. Agents browse, compare and buy; clicks stop meaning what they meant; the shelf turns into an API. Every CMO and CCO we talk to is asking the same two questions –&nbsp; am I still visible, and can I still prove what worked – and almost nobody can answer them.</p> <p><strong>What you will be doing</strong></p> <p>You will own the hardest question in marketing: what actually caused the sales. Marketing Mix Modelling, incrementality testing, geo-lift, attribution –&nbsp; for brands where a three percent reallocation of budget is worth more than most consulting engagements.</p> <p>And the question is getting stranger. Alongside the media work, you will sit close to three things we build: synthetic brand personas, designed end to end from character bible and voice through to the generative video and avatar pipeline that produces them, running in the client's own cloud with human authorship and copyright chain documented; AI creative studios that generate brand-consistent product imagery and video at retail scale; and trend intelligence engines that detect a movement, predict where it goes, and explain why it matters commercially – early enough to change what a business actually makes.</p> <p>None of this has a settled measurement stack. Proving it works is the same discipline as proving a television budget worked, applied to something nobody has measured before. That is also you. You will do it as a hands-on scientist and as a trusted advisor to the CMO – both, not either.</p> <p><strong>What that looks like day to day:</strong></p> <ul> <li><strong>Direct access to the frontier labs.</strong> We work with Anthropic, Google, OpenAI and Mistral as partners, not vendors on a procurement list –&nbsp; early access, joint bets, and the freedom to build with what shipped this quarter rather than what was safe two years ago.</li> <li><strong>You own the outcome, not the deliverable.</strong> We are in the client's environment from day zero and still there after launch. No prototype handoffs. When something breaks in production, you are close enough to see it and fix it.</li> <li><strong>Problems worth your brain.</strong> Bayesian MMM at group and geo level. Calibrating models against live experiments. Privacy-constrained measurement where the easy signal has disappeared. Working through identification assumptions with people who care as much as you do about getting them right.</li> <li><strong>A direct line to the CMO.</strong> Not a workstream lead reporting up through three layers. You present the method, you defend it, you own the recommendation – and you're in the room during qualification, before anything is signed.</li> <li><strong>Room to shape the offer.</strong> Measurement is not a finished product. You will help decide what we productise next, what we stop doing, and how we take it to market.</li> <li><strong>People who make you better.</strong> You will work alongside scientists, engineers, creatives and strategists who will pick your models apart and then help you fix it. Nobody here ships alone, and nobody is left to fail quietly.</li> </ul> <p><strong>What We Are Looking For</strong></p> <p>Seven things. The first four are non-negotiable –&nbsp; if you don't tick every box but the problem interests you, write to us anyway.</p> <ul> <li><strong>Deep marketing measurement expertise.</strong> Around eight years of building, not just presenting, MMM, incrementality and attribution work, or the depth to match. You have shipped models that changed how a real budget was spent, and you can talk about the ones that failed.</li> <li><strong>Genuine marketing fluency.</strong> You know how retail media behaves differently from paid social, why brand-building resists short-term measurement, and what a media planner is actually going to do with your output on Monday morning. Measurement without marketing intuition produces confident nonsense.</li> <li><strong>Serious data science.</strong> Causal inference and Bayesian methods as a working craft, not vocabulary. Comfortable across the wider toolkit – forecasting, optimisation, applied GenAI – and still writing code. You would rather debug it than delegate it.</li> <li><strong>CMO-grade communication.</strong> You can hold a technical audience for an hour on the identification strategy and a CMO for ninety seconds on what to do about it, and be trusted by both. This includes saying "the result is inconclusive" when it is.</li> <li><strong>Production instincts.</strong> MLOps and cloud industrialisation. A model living in a notebook is a hobby, not a deliverable.</li> <li><strong>Presence before the contract.</strong> You are in the room during qualification and solution design, not handed a signed scope. Buyers judge us on the calibre of the person in front of them –&nbsp; often that is the whole evaluation.</li> <li><strong>AI-native ways of working.</strong> You use AI on your own work daily. If you would rather keep the ways of working you already know, ours will frustrate you.</li> </ul> <p>You'll thrive here if you'd rather be wrong in production than right i