Head of Data & AI
Own the data layer across ForLab+ and Link: data quality, forecasting accuracy, and the learning loops that make both products more useful over time. This is a practitioner role, not research.
About the role
Your job is to make Opian's products smarter every time they're used. ForLab+ learns from facility-level supply chain data to improve forecast accuracy; Link learns from clinical outcomes to sharpen decision support. Both run on messy, real-world data from across Sub-Saharan Africa — incomplete lab reports, handwritten patient records, network outages, data entry errors.
You'll build systems that improve data quality at the source, establish accuracy targets, and create feedback loops that make forecasts and decision support demonstrably more useful. You'll work with the CTO on data pipelines and the VP Delivery on how facilities generate and send data.
This is not a research role. You publish internally first. You care about precision recall tradeoffs, model calibration, and whether a clinician will trust a recommendation. You can explain models to Ministry of Health officials who ask 'How do you know this is right?'
What you will own
- Data quality: establish standards, build monitoring, flag bad data early, create incentives for clean input
- Forecasting accuracy: own the loss function, experiment with models, ship improvements that move the needle from ±7.1% baseline
- Decision support: Link's clinical recommendations must be calibrated, explainable, and reduce clinician second-guessing
- Data publishing and external validation: Opian publishes forecasts and outcomes publicly; you ensure accuracy and interpretability
- Learning infrastructure: feedback loops that make both products improve as they run at scale
Year one goals
- Improve national forecast accuracy from ±7.1% to ±5% through data quality work and model refinement
- Build an internal data-quality scoring system for incoming facility data; flag low-quality submissions before they corrupt forecasts
- Ship a forecasting copilot feature into ForLab+ that helps planners interpret predictions and adjust for local context
- Establish Opian's data publishing standards: what we publish, how we validate, how we respond to external questions
What we are looking for
- Applied data science background: ML courses in school, then real-world data work. 5+ years shipping systems where model performance matters operationally.
- Comfortable with messy real-world data. You've worked with incomplete datasets, outliers, data entry errors. You're good at exploratory analysis and root-causing quality drops.
- Can explain models to non-technical audiences. You can draw a diagram that makes a Ministry official understand why a forecast is what it is.
- Pragmatist about ML: you know when a simple heuristic beats a complex model, and you're not precious about method.
- Interested in health: you read the clinical literature, you understand why a drug stock-out or diagnostic delay matters, you care about the people using your systems.
How we interview
Initial call
A 30-minute conversation with someone from the team — usually the role's direct manager. We talk about your background, what you're looking for, and why Opian. We're looking for genuine interest in health systems and a communication style that works for us.
Craft interview or case
Depending on the role, this might be a take-home assignment (CTO candidates do a system design; data candidates might analyze a dataset) or a live conversation around a case study from our work. We're not testing trivia — we want to see how you think through ambiguity.
Team conversations
You'll talk with 2-3 people you'd work closely with: the CTO, Head of Data, or VP Delivery. These are working conversations, not interrogations. We talk about how we ship, how we disagree, how we debug.
Offer
If we're both excited, we make an offer. Compensation is competitive, commensurate with experience, plus meaningful equity and relocation support where relevant.
About Opian
Opian builds decision systems for health supply chains and clinical care. ForLab+ helps countries forecast drug, vaccine, and test demand across thousands of facilities. Link is a clinical decision support system for hospitals. Both are deployed at national scale in Ethiopia and expanding across Sub-Saharan Africa.
We're a 15-person team based in Addis Ababa. We're funded by the Gates Foundation. We move quietly and ship. We believe in writing things down, talking to users, and making decisions with authors, not committees.
More about Opian