Architected and developed the frontend for Remotify- a Phillipines based Employer-of-Record platform using a monorepo structure to streamline development and maintenance. Worked on important features on the frontend - implemented shared auth, client onboarding workflows, and addition of employees. Occasionally worked on backend services in Golang to support these features.

Built an interactive chatbot which would help you get to know more about me! It's built using Next.js entirely, utilizing the Vecel AI SDK to connect to OpenAI's GPT-4 model for generating responses. Built mainly to understand the concepts of RAG and chunking in a better way by experimenting with different techniques to improve response accuracy and relevance.

Developed the frontend for a vertical search engine designed to enhance agency discovery with advanced filtering by country, industry, and company size. I implemented real-time facet handling, ensuring that filter counts updated instantly as users applied or removed criteria. A key challenge was guaranteeing that the search page was always server-side rendered, delivering fast load times and optimal SEO performance. This project strengthened my understanding of Server side rendering, and pushed me to deliver a seamless user experience to work on applications with instant data updates.

Built for a startup in the US - GovGPT. It is a Generative AI threat detection system with a Next.js frontend, Python backend and WebRTC streaming. Frontend sends the stream and receives bounding boxes over websockets in real-time. Leveraged the power of WebSockets and WebRTC to ensure that the video stream and bounding boxes were in sync with low latency. The text based response generation is handled on the backend and exposed via a FastAPI endpoint which internally makes calls to OpenAI's API.