Delivered work Absolute tenure: October 2021–November 2024
Rainfall data and insurance
The starting rainfall data used a 25 km grid. My product work connected a finer 10 km × 10 km model with validation and an actuarial pricing task.
My role Rainfall-data requirements, ground-truth validation and actuarial collaboration.
Why this mattered
A rainfall grid can group nearby locations even when their conditions differ. More local information would matter only if it was valid and usable for the insurance-pricing decision.
A finer grid for a pricing requirement
The starting constraint was publicly available 25 km rainfall information and the cost of obtaining more detail. The project developed a 10 km by 10 km model using satellite estimates, local weather stations, terrain and historical rainfall patterns.
I worked on the product use of that capability. The requirement joined a finer spatial unit with the information an actuarial team needed for premium tables, rather than end with completion of a model.
Ground-truth work across monsoon cycles formed part of the account, alongside actuarial collaboration. The product handoff connected the rainfall output to a specific institutional pricing task.
Grid use and sandbox are different milestones
The historical account describes grid development and actuarial pricing use. It separately describes the related PMFBY sandbox as continuing into execution. That stage stays distinct from a completed insurance rollout.
Moving from 25 km to 10 km specifies spatial resolution, not accuracy. In another data product I would ask what local distinction the receiving team needs, how that distinction is validated and what changes in its decision.
FinStac’s later current catalogue is retained only as background. Its cover and distribution propositions do not establish the historical grid’s model performance, policy issuance or claims outcome.
Grid scope and reported use
Documented scope: the rainfall grid changed from 25 km to 10 km × 10 km. This describes spatial resolution, not model accuracy.
An earlier career record reports a 27% premium reduction for a use case, without its cohort, comparison or period. The related PMFBY sandbox is recorded as ongoing execution, not completed rollout.
Sources & technical detail
Original artifacts, detailed tables and source notes.