Senior Data Scientist

We're trying to predict how consumers will respond to ads and product concepts before companies spend millions producing them. We’re aiming to go well beyond simple LLM wrappers into actual predictive models trained on real survey data, validated against real market outcomes.

It's a significant challenge. The data is multi-modal and mixed qualitative and quantitative. The validation requirements are strict. Many of our competitors make vague AI claims with no published accuracy metrics. We'd rather be honest about what works and what doesn't.

We need someone who can push the research forward.

You'd own the modelling approach for one of three domains — Advertising (predicting ad effectiveness), Innovation (forecasting product success), or Tracking (brand health dynamics).

You won't just be handed a dataset and told to find patterns. We're significantly investing in custom-designed surveys built specifically to train these models. You'd be shaping what data we collect, not just how we interpret it.

That means thinking through questions like:
What stimulus variations do we need to capture the full distribution of ad quality? How do we ensure proper category and demographic balance? What response patterns actually predict market outcomes? The data asset itself is part of what you're building.

Some of the harder modelling problems:

Validation frameworks. We need to know exactly how well our models perform, and more importantly, where they fail. You'd design how we measure this.

Respondent-level synthesis. Most synthetic approaches just predict top-level scores. We're trying to generate full synthetic respondent profiles — hundreds of variables per individual. Architecturally quite different from regression to a mean.

Cold start. How do you predict response to a genuinely novel product category or ad format? Where can we use transfer learning?

Multimodal inputs. Video, images, text, brand context — figuring out what features actually matter for predicting human response.

PhD or equivalent depth in ML, statistics, computational social science, or similar. We want someone who's driven research, not just applied existing tools.

Evidence you can solve problems that don't have existing solutions. Papers, repos, production systems — something that shows how you think.

Interest in experimental design, not just model fitting. If thinking about what data to collect is as interesting to you as what to do with it once you have it, that's a good sign.

Tolerance for messy data. If imperfect real-world data frustrates you more than it energises you, this probably isn't the right role.

Strong fundamentals over framework knowledge. We care more about whether you understand why approaches work than which libraries you've used.

Ability to communicate across disciplines.

You'll work with research scientists who think in survey methodology terms. Translation skills matter.

Budget to build the dataset properly. We're not trying to do this on the cheap. Custom data collection designed specifically for training predictive models.

Historical data to learn from. Years of survey responses that inform what works — even if the pure training data needs to be purpose-built.

Infrastructure that doesn't fight you. We've spent a lot of time building a data platform that lets data scientists actually test ideas quickly.

Domain experts. Research scientists with deep expertise in survey methodology and market research. You won't be figuring out the domain alone.

• Three company-paid mental health days of rest per year (these are pre-scheduled, so the entire company can take the same days off regularly to reset)
• Generous PTO policy - Minimum of 20 days of paid time off per year, plus national holidays. Leave over 30 days requires manager and People team approval.
• Thoughtfully designed offices to support both individual work and collaboration without interrupting others
• Private medical healthcare cover
• Medical Aid
• Group risk, life & disability contributions
• Wellbeing benefits such as free yoga and access to trained therapists / counsellors
• Paid 24h secure parking in Cape Town
• Free coffee,…