* Contributed to an AI-powered web app focused on revamping the online retail experience (full-stack). Worked on product-specific inquiries and the parsing of retailer data into our database (React, Elixir)

* Explored and prototyped online purchasing through an AI agent. Efforts included site navigation from natural‑language queries, product selection verification, and the integration of payment providers' solutions to securely authenticate agent-initiated purchases (Elixir, TypeScript, Python)

* SPC x Anthropic Hackathon 2025 finalist: built an AI storefront providing a chat-based online shopping experience. Implemented product Q&A post-search and worked on AI prompting across the app (Python)

* Designed a database schema for a universal product catalog, aggregating diverse data sources (PostgreSQL)

* Implemented the app layer of the AI Pin’s on-device Capture Experience, involving a Laser Ink UI, AI agent action routing, and data persistence (Kotlin, Java)

* Built on my internship work to ship the AI Pin’s video feature by iterating on the capture architecture (removing thermal constraints on video length), working with the device’s sensors to enable server-side video stabilization, and following up on beta customers’ feedback (Kotlin, Java, C++)

* Worked on the development of the AI Pin’s Camera HAL wrapper, providing an entry point for any on-device experience to open a stream or capture a frame, and abstracting away some AOSP complexity (C++)

* Implemented the first architecture of the AI Pin’s photos+videos cloud syncing logic and helped with bug fixing and architectural decisions after handing it off (Java)

* Kicked off a cross-functional effort for CosmOS’s Health Experience. Built on already-existing food recognition work to build a demo focused on wellbeing. From there, assisted in the creation of a bigger team encompassing several health-facing features. In this team, I helped bring the AI Pin’s step counter from demo to a shippable state, and worked on the health AI agent and cloud syncing mechanisms for health data (Kotlin)

* Picked up the “Double tap to capture” feature by turning a prototype implementation into a more robust architecture in its separate process and adding retry/error/persistence logic (Java)

* Built the first implementation of video on the AI Pin working with Android Camera API’s and an early Humane camera architecture, providing cleaner recording API’s at the app layer and building on the existing photo architecture to account for a new type in non-capture logic (Java, C++)

* Implemented the first iteration of a Laser Ink UI for the AI Pin’s Capture Experience, including a prototype of a view finder on the palm and an object recognition UI (Java)

* Implemented software used for research about the decision-making process

* Eye-tracking data was collected from a patient’s web cam and analyzed to make inferences about the effect of visual stimuli on choices