I'm a Computer Science student at The University of Texas at Austin (Expected May 2027). I'm passionate about building full-stack applications and AI-powered tools that solve real-world problems. Currently working at Carpaal on frontend Design.
-
Gist- a Natural Languge File Exploerer: Building a NLP file exploerer using embeedings to search for files with discriptions instead of filename
- Using Rust, Electron, and Ollama to create a priavte AI powered sematic search engine.
- Allows for file exploerer features like previewing and opening files, adding and delting files, and asking file speific questions to locally run AI
-
Cognify - An AI powered Study Assistant: Developing an accessibility-first application that converts complex educational content into neuro-friendly formats.
- Built with Next.js 14, TypeScript, and Tailwind CSS to deliver features like bionic reading and gamified chunking.
- Engineered a custom Multi-Provider AI Mesh (utilizing Groq, SiliconFlow, and Google Gemini) that intelligently routes prompts based on complexity to maximize free-tier throughput.
| Category | Technologies |
|---|---|
| Languages | Java, Python, C/C#, JavaScript, SQL, HTML, CSS |
| Frameworks & Libraries | PyTorch, Django, React, Node, Next.js, Pandas |
| Developer Tools | Git, AWS, REST APIs, Docker, CI/CD |
(Python, PyTorch, FastAPI, React, Docker, AWS S3)
- Developed and deployed an end-to-end deepfake detection web application to reduce manual verification time.
- Trained and fine-tuned a lightweight EfficientNet-B0 model for binary classification, achieving 94% validation accuracy through data augmentation and early stopping.
- Engineered a scalable REST API using FastAPI and Docker, achieving an average inference latency of <500 ms per image on Hugging Face Spaces.
- Built a responsive user interface with React on Vercel for intuitive image uploads and clear result visualization.
(React, TypeScript, FastAPI, Pygame, Docker)
- Architected a full-stack application during the 24-hour HackTX 2025 hackathon that generates fully playable 2D games from natural language prompts.
- Developed a specialized 'Sequential Chain of Command' prompting model that achieved a 92% success rate in parsing complex user prompts into valid, engine-readable JSON.
- Engineered critical FastAPI endpoints to connect the React/TypeScript frontend to the Pygame game engine, enabling a seamless "prompt-to-game" user pipeline packaged with Docker.
(Next.js, Django, Stripe API)
- Owned and architected the end-to-end full-stack rebuild of the nonprofit's website, using Next.js and Django.
- Established a secure, multi-provider payment system by integrating the Stripe API.
- Led a hosting migration projected to cut annual maintenance costs by 66%.
- Streamlined the user donation flow by redesigning the UI/UX with responsive components and implementing SEO best practices.
- Email: [email protected]