Synapse

Built in 24 hours at NwHacks 2024, Synapse is an AI-powered tool that transforms dense PDF documents into interactive, easy-to-understand flowcharts. Working in a team of two, we leveraged Large Language Models (LLMs) and serverless architecture to automate the entire process, from content extraction to diagram generation.

I focused on building the backend and designing the UI. The backend was architected using AWS Lambda, the Serverless Framework, and S3 for scalable, secure file storage. This setup allowed for asynchronous processing of API calls. For the frontend, I used TypeScript and Mermaid.js to render the dynamic flowcharts from the data processed by the AI. A major challenge was optimizing the data flow between our microservices to ensure real-time rendering.

The final product could successfully accept a PDF upload, process its content using an AI model, and render a dynamic, interactive flowchart. The tool was praised by judges for its technical sophistication and practical application in simplifying complex information.

This hackathon was an intense exercise in rapid prototyping and creative problem-solving with AI. It sharpened my skills in asynchronous programming and API integration under tight deadlines. Collaborating closely with a partner also reinforced the importance of clear communication and effective teamwork in a fast-paced environment.