Human-Centered AI for Public Systems

Census-scale reading assessment for state education systems

Led 0→1 design for an AI-based reading assessment tool, translating field insights into the features that drove adoption by SCERT Rajasthan and the Vidya Samiksha Kendra in Gujarat — enabling statewide, census-scale assessment. Deployed through G-Shala in Gujarat since 2023 and across Rajasthan government schools since 2025, the tool is now used by roughly 30,000 public schools and is scaling nationally across ten Indian languages.

An LLM health assistant for frontline Anganwadi workers

Led design and user research for a low-literacy-optimised LLM assistant, piloted across six districts and now scaling pan-India through Poshan Tracker with the Ministry of Women and Child Development. Selected as a real-world impact use case at the India AI Impact Summit. A multimodal, multilingual assistant built with the Ministry of Health and Family Welfare and the Ministry of Women and Child Development, supported by Google.org.

Diabetic retinopathy screening, co-designed with AIIMS Delhi

Led design research and co-designed a diabetic retinopathy screening tool that supports optometrists, nurses and community health officers in grading retinal images to international benchmarks and triaging referable cases. It has flagged over 3,000 referable cases and more than 1,000 instances of vision-threatening retinopathy for urgent care.

Winner of the AI for ALL: Global Impact Challenge at the India AI Impact Summit 2026, selected from over 4,600 applications across more than 60 countries.

A farming assistant answering queries through Kisan Call Centers

Led user research and 0→1 product design for an LLM-based farming assistant deployed across thirteen states via Kisan Call Centers, giving advisors instant IMD weather guidance as concise, shareable summaries in eleven Indic languages.

An immunization tracker a parent can read at a glance

Infants in low-resource settings often have no reliable immunization record, and formal tracking systems assume a literacy and paperwork burden many families cannot carry. Shishu, Sui aur Dhaaga — “child, needle and thread” — borrows the form of the nazar thread already tied on infants across India for protection, reframing invisible vaccine-preventable disease as the threat it guards against.

The insight came from the field: among migrant and floating populations, parents already understood preventive protection through nazar — guarding a child against unseen harm. Vaccine-preventable disease is unseen harm. The design didn’t introduce a new idea; it attached to one that was already there.

A health worker adds one bead at each of the eight required visits. Parents and nurses can read a child’s vaccination progress instantly, without records or literacy, and the artefact stays on the child rather than in a folder. The team proposed distribution through hospitals at birth, at under ₹10 per bracelet.

With Kavya Murugan, mentored by Vineeta Rath. Winner, Stanford Center on Longevity Design Challenge 2019–20 — Closing the Inequity Gap: Designing for Affordability. Also shortlisted in the Gates Foundation / OpenIDEO Immunization Innovation Challenge.