My Projects
A comprehensive showcase of my work in AI, Machine Learning, and Software Development, spanning healthcare, gaming, and innovative applications.
Kolate.ai Chat Copilot
GenAI Project
Advanced chatbot for an AI research pharma startup featuring a complex pipeline with multiple functionalities including RAG for clinical trial data, data analytics via LLM-based code interpreter, and generic medical query retrieval.
Key Features:
- RAG & RAFT implementation for clinical trial data
- LLM-based code interpreter for data analytics
- Medical query retrieval using PubMed integration
- Multi-model support (GPT-4o, Llama3 70B, Mistral 8x7B)
Clinical Trial Prediction Engine
ML Project
Predictive analytics system for clinical trials, designed to predict different aspects of trials for new drugs or vaccines before and during the trial phase, with focus on synthetic feature generation.
Technical Highlights:
- Advanced regression and classification models
- Synthetic feature generation methodology
- Comprehensive data preprocessing pipeline
- Hyperparameter tuning and data augmentation
Diabetic Retinopathy Detection
Computer Vision
Advanced computer vision project aimed at detecting different levels of blindness in retina images for diabetic retinopathy diagnosis. Experiments with various deep learning models from simple CNNs to complex architectures.
Technical Implementation:
- Low to high-end model experimentation
- Low resolution image optimization
- CNN and advanced architecture comparison
- Medical image preprocessing pipeline
Movie Rating Generator
NLP Project
AI-based movie rating system using sentiment analysis with fine-tuned Twitter roBERTa model. Features a statistical formula to convert sentiment scores into meaningful movie ratings.
Project Features:
- Fine-tuned Twitter roBERTa-base model
- Custom statistical rating formula
- Flask web application interface
- OMDB API integration
Text Summarization App
NLP Application
Web-based text summarization application using fine-tuned T5 small model. Provides efficient text summarization capabilities through a user-friendly Flask interface.
Technical Stack:
- Fine-tuned T5 small model
- Flask web framework
- Real-time text processing
- Responsive web interface
EdTech AI Assistant
Full-Stack GenAI
Full-stack LLM-based chatbot designed for an EdTech company (USDC Global). Built using only open-source models and resources for cost-effectiveness.
Development Approach:
- Complete full-stack development
- Open-source model implementation
- Cost-effective architecture
- Data scraping and RAG implementation
Additional Projects & Research
Financial News Impact Analysis
NLP project analyzing financial news headlines and their market impact using HuggingFace Transformers.
Document OCR & Classification
Extraction and classification system for different document types using Tesseract OCR and fine-tuned GPT NEO 2.7B.
Parallel Matrix Multiplication
High-performance matrix multiplication using Strassen-Winograd algorithm with CUDA-C and parallel programming.
CIFAR-10 Classification Study
Comprehensive comparison of neural network architectures from simple CNNs to Li-Net based models on CIFAR-10.
Gaming ROAS Forecasting
Time series forecasting models for Return on Ad Spend (ROAS) predictions in the gaming industry with real-time deployment.
GenAI Playground
Advanced GenAI playground using GPT-4.1, Llama 3.3 70B, Llama 4, and TTS models from GCP and ElevenLabs.
Interested in Collaboration?
I'm always excited to work on innovative AI and ML projects. Let's discuss how we can create something amazing together.