Built at ConUHacks 2026 The future of search isn't keywords; it’s semantics.

Traditional lost and found systems are fundamentally broken. They rely on manual spreadsheets or public lists that compromise security. Publicly listing a "Gold Rolex" invites false claims, while not listing it prevents recovery. findIt bridges this gap using a "blind matching" architecture that prioritizes privacy and security.

findIt automates the "handshake" between a lost report and a found item using high-dimensional mathematics. By leveraging Vector Embeddings, the system matches items based on their semantic meaning rather than exact word matches, solving the friction of manual searching in large-scale environments.

1. Multimodal Ingestion: When a user uploads a photo, it is processed by a BLIP Processor to generate a descriptive textual representation.
2. Embedding Generation: This description is passed to the Google Gemini API to generate a high-dimensional vector embedding.
3. Vector Storage: Embeddings are stored in MongoDB Atlas, which acts as our primary source of truth.
4. Semantic Retrieval: To find a match, the system performs a Vector Search using Cosine Similarity between the user's inquiry vector and the inventory vectors.

• Frontend: Angular (Reactive UI/UX)
• Backend: FastAPI (High-concurrency & Asynchronous processing)
• Database: MongoDB Atlas (Vector Search & NoSQL storage)
• AI/ML: Google Gemini API, BLIP Processor, Python

• Lifecycle Management: Managing the lifecycle of vector embeddings from generation to search-indexing.
• Mathematical Search: Leveraging MongoDB’s search indexes to perform complex math (Cosine Similarity) on thousands of items in milliseconds.
• Human-Centric Design: Designing software that solves an emotional human problem (losing valuables) with a balance of empathy and technical security.