Developed a dynamic SQL playground enabling users to write and test queries in a mock environment, with real-time results, execution metrics, and timestamped logs to simulate database performance.

Implemented robust input validation and server-side middleware with custom error handling to ensure reliability and help users explore query optimization strategies safely.

Implemented global UI state management using Zustand’s create, devtools, and subscribeWithSelector middleware to reduce prop drilling and enable performant, observable layout updates across views.

Built reusable React components to boost layout flexibility and speed up development, leveraging its declarative rendering, lifecycle control, and smooth integration with Zustand and Tailwind for scalable UI architecture.

Adopted TypeScript across the codebase, building reusable interfaces and shared types to standardize API and component contracts. Configured strict compiler options to reduce technical debt and enable faster, safer team development.

Used Docker to package the app into a single image, deploying via AWS Elastic Beanstalk on EC2 for consistent cross-environment performance, enabling secure certificate management with Amazon S3 to ensure safe deployment.

Built a modular Express.js backend with centralized error handling and service-based integrations for OpenAI and Pinecone, structured for scalability and future feature expansion.

Enabled fast semantic search and recipe retrieval based on user-provided ingredients by embedding a cleaned recipe-5k dataset using OpenAI’s embedding model and uploading it to Pinecone.

Integrated Pinecone to store and semantically search embedding vectors, leveraging upsert/query endpoints for real-time indexing and fast retrieval with OpenAI embeddings.

Implemented a dual-model pipeline using OpenAI’s embedding and GPT-4o-mini models to match ingredient inputs to recipes and generate coherent, human-readable suggestions.

Built dynamic reward tracking and redemption features, allowing users to manage multiple merchant offers from a unified dashboard; implemented input validation and secure API design to protect sensitive user actions.

Designed a relational PostgreSQL schema with one-to-many (merchants → users) and many-to-many (users ↔ rewards) relationships, normalized for scalability and integrated via RESTful API routes.

Built a role-based login system with Node.js and PostgreSQL, validating stored credentials and dynamically serving different views to enforce access control and tailor user experience.