THE ROLE
We're making a founding hire: a Staff Data Scientist in Marketing Science to own the mathematics behind our paid growth - the causal measurement, the response curves, the customer-value models and the optimisation that decide where our multi-million-dollar, always-on acquisition budget goes across Google, Apple Search Ads, OpenAI ads, Meta, TikTok and X. There is no inherited methodology - you design it from first principles.
Your job is to answer one question, continuously and with evidence: where should the next $1 million go to generate the most additional profit - and how confident are we?
This is a hands-on, builder role. You'll write the models yourself, design the experiments that keep them honest, and sit next to the people moving the budget. Our data engineers own the pipelines; you own the science.
WHAT YOU'LL DO
• Build our marketing measurement system - a Bayesian marketing mix model with adstock and saturation, channel-level response curves and full uncertainty, in PyMC, Stan, NumPyro, Meridian or a framework you can defend
• Separate attributed from incremental - design and analyse geo-experiments, conversion-lift tests, holdouts and synthetic-control studies, and use the results to calibrate the model
• Turn models into budget decisions - estimate marginal ROAS and marginal CAC by channel, run constrained budget optimisation and give the paid acquisition team a clear, risk-aware allocation every planning cycle
• Model what a user is worth - build LTV, retention and payback models tied to activation and trading activity, so we optimise for profit, not platform-reported ROAS
• Measure what the platforms can't - build defensible approaches for newer or opaque channels (OpenAI ads, X, Apple Search Ads, crypto-native networks) where lift tooling is thin or missing
• Work alongside the Head of Paid Acquisition & Performance Marketing to set test roadmaps, define success metrics and decide what to scale and what to stop
• Ship production-quality code - versioned, tested and reproducible - and partner with Data Engineering on the inputs the models need
• Explain uncertainty plainly - write decision memos that founders and operators can act on, including when the right answer is "we don't know yet - here's the test"
WHAT WE'RE LOOKING FOR
• Advanced quantitative training - a PhD, MSc or equivalent depth in mathematics, statistics, econometrics, physics or operations research. What matters is that you can derive it, not just run it
• 4+ years of hands-on experience (PhD research counts), with a track record building marketing mix or advertising response models that moved real budgets - in-house at a consumer, fintech or app business, or at a measurement company (Recast, Haus, PyMC Labs, Google, Meta or similar) - with results you can explain and evidence
• Hands-on causal inference - you have designed, run and analysed incrementality experiments, and you know what to do when the experiment and the model disagree• Deep Bayesian modelling - hierarchical models, priors you can justify, identification, posterior diagnostics and the computational know-how to make them fit at speed
• Strong Python or R in production - you write the model code yourself
• Commercial judgment - you think natively in incremental profit, LTV, CAC and payback, and you can turn a posterior into a recommendation a non-specialist will act on
• A scrappy, high-energy operator mindset - you thrive in a fast-paced, resource-lean environment where you're building the playbook as you go
NICE TO HAVE
• Contributions to open-source measurement tools - PyMC-Marketing, Meridian, Robyn, LightweightMMM, CausalPy, Stan, GeoLift or similar
• Mobile app user acquisition - SKAdNetwork, AppsFlyer or Adjust, and Apple Search Ads
• Crypto experience - as a retail trader, or through work at a crypto startup or larger crypto company
Sophisticated equations matter. Better allocation decisions matter more.
HOW WE'LL MEASURE SUCCESS
• First 30 days: Current measurement audited, a baseline response model live for core channels, and the first incrementality tests designed with the paid acquisition team
• First 90 days: An experiment-calibrated model producing channel response curves, marginal returns and a recommended allocation with confidence ranges, used in the budget cycle
• First 180 days: Budget reallocations driven by the model delivering measurable incremental profit, with an always-on test-and-learn programme across existing and new channels
COMPENSATION & BENEFITS
• Base salary: $200,000 - $300,000, depending on experience and scope
• Performance-related compensation: Generous performance bonus; equity or token participation depending on level
• Benefits and relocation support: To be confirmed for this role
LOCATION & ELIGIBILITY
• New York preferred, working in person with our growth team. For exceptional candidates we will consider remote work with strong overlap with US Eastern hours and regular time in New York.
• For New York-based roles you must be authorized to work in the United States. Please tell us in your application whether you will now or in the future require visa sponsorship.
HOW TO APPLY
Send your résumé and a short example of a model or experiment that changed how advertising budget was spent: the question, your approach, and what moved. Links to papers, talks or code are welcome. No cover letter needed.
careers@batoncorporation.com
If you'd rather move the budget than write the attribution report about it - this is your role.
