Deepraj Parab

I've built AI-driven applications with Python, FastAPI, PostgreSQL, and machine learning workflows, including sales conversion prediction services at WEBiSOFTECH.

During my AI/ML internship, I developed REST APIs for audio processing and prediction, connecting speech-to-text, NLP feature extraction, model inference, and predictive analytics into a unified workflow. I also contributed to debugging, testing, optimization, and deployment of AI applications.

In my AI Physiotherapy Assistant project, I created real-time posture analysis and joint-angle tracking using OpenCV and MediaPipe, with FastAPI services for authentication, exercise tracking, and analytics.

I've also architected an AI expense tracker using Multi-Agent AI and RAG, integrating Tesseract OCR, Gemini Vision, Pinecone, fraud detection, categorization, reminders, and recommendations.