Projects
Ashowcaseofmytechnicalprojects,hackathonwins,anddevelopmentwork.
Project
Sign1o
CNN model for sign language recognition with real-time processing capabilities.
Achievement
AI-powered sign language recognition system
Key Features
- Built a CNN model (Python, TensorFlow) for sign language recognition, achieving 90%+ accuracy on 100+ video samples with a pipeline for 5% quarterly accuracy gains.
- Enhanced a NestJS backend service using Prisma ORM with a PostgreSQL database by integrating Prisma Accelerate; reduced query latency by 70% and connected backend with frontend using WebSockets to serve 50+ signs.
Technologies
PrismaPostgreSQLNestJSNext.jsShadcnFastAPIPyTorchGoogle Gemini
Project
SafeGuard
Convolutional neural network for skin cancer classification with containerized deployment.
Achievement
Medical AI diagnostic tool
Key Features
- Built a convolutional neural network using TensorFlow for skin cancer classification, achieving 52% accuracy.
- Architected and deployed backend microservices with Flask, Express.js, and MongoDB, containerized via Docker, and automated deployments with Azure CI/CD pipelines, improving delivery speed and system reliability by 187%.
Technologies
DockerAzureReact.jsNode.jsFlaskTensorFlow.js
Project
FoiceDetect
Deepfake voice detection model using SVM classifier with high accuracy detection.
Achievement
Nosu AI Hackathon Winner - $325
Key Features
- Built deepfake voice detection model using SVM classifier on 100+ audio samples, achieving 94% accuracy.
- Enabled real-time voice detection to present to 150+ AI professionals at USF ACM AI Convention.
Technologies
NumPyScikit-learnLibrosaDjangoNext.js