Data Science & Public Health Intern - #1170743

Company: Covee (BDV OPCO PTE. LTD.)

Location: Remote/ Hybrid.

Duration: 3 to 6 months. Full time or Part time.

Eligibility: Singapore Citizens and PRs, or international students holding a valid Student’s Pass from an institution approved by MOM.

To apply: CV. Include a copy of a data analysis presentation or artefact you have produced in the past. If you have written something you are proud of, send that instead of a cover letter.

The role

Covee builds software for employee health benefits. Today a broker gets a spreadsheet of last year’s claims from the insurer and reports back on what happened. We are building the part that looks forward: what a company will likely spend next year, who is probably unwell but has not claimed yet, and which health programmes are worth paying for. This role is the predictive side of that, and you would work directly with the founders, who have built and scaled venture-backed companies and brings operating experience from places like Meta, Blackrock, Wish and Mercer.

What you will work on

• Research the medical and health economics literature to find which health programmes genuinely reduce risk, and by how much

• Build the models that turn those findings into money: how many staff are eligible, how many will take part, how much their risk drops, and what that avoids in cost

• Predict what a company will spend next year, as a range with the reasoning attached rather than a single confident number, and test thoseforecasts against historical outcomes where data allows

• Find the risk hiding in a company’s population, by comparing the illness showing up in claims against national rates for its age and sex mix

• Get claims data into shape to support all of the above, since every insurer sends it differently

What we are looking for

• A degree or diploma with a quantitative health focus: public health, biostatistics, epidemiology, health economics, bioinformatics, statistics, data science or actuarial science
• Hands-on experience in R or Python, including at least one relevantproject in healthcare analytics, population health modelling or claims forecasting. University research and internship projects count. We would like tosee what you built, how you validated it and what you learned.
• Real analysis experience in R or Python, either is fine
• Statistics you can explain and defend when questioned, not just run
• Enough SQL to answer your own questions
• Care with incomplete and inconsistent data. Some claims have no diagnosis recorded, and no two insurers format anything the same way

How we use numbers

We do not claim numbers the evidence cannot support. Screening often looks like it raises cost in year one, because finding people who are unwell means paying to treat them, and we say so. We do not credit a programme with savings the studies do not show, and we will not put a dollar figure on an outcome that was never measured that way. Every number links back to a published paper and stays marked draft until a doctor or health economist signs it off.

The same honesty applies to scope. With a few thousand insured people across a few dozen companies, you can say what a year cost, where the spending concentrates, and roughly what next year looks like. You cannot predict which individual person will become expensive, whichever algorithm you use. Part of the job is knowing the difference and saying so out loud.