Selected Projects with Real-World Impact

Curated Hugging Face Model Hub featuring fine-tuned transformers, custom datasets, and inference API wrappers. Demonstrates model versioning, evaluation metrics, and production deployment …

Hugging Face, Transformers, PyTorch, NLP, Model Hub, MLOps

Retrieval-Augmented Generation (RAG) system for semantic search over PDFs and documents. Uses FAISS vector embeddings, LLM-powered answering, document parsing, and web UI …

Python, FAISS, Vector Embeddings, LLMs, NLP, Streamlit, RAG

Advanced interactive dashboards with drill-down analytics, KPIs, real-time data feeds and exportable reports. Built with Plotly, Streamlit, and optimized Pandas pipelines for …

Plotly, Streamlit, Pandas, SQL, JavaScript, Data Analysis

Production-grade ML API built with Django for tip prediction. Features model versioning, REST API endpoints, Docker containerization, and scalable inference pipelines for …

Django, DRF, Scikit-learn, PostgreSQL, Docker, REST APIs

Efficient LoRA-based fine-tuning framework for large language models. Includes training scripts, evaluation pipelines, hyperparameter optimization, and Hugging Face deployment examples for production …

LoRA, Transformers, PyTorch, Hugging Face, Fine-tuning, QLoRA

Offline-first AI chatbot powered by local LLM models via Ollama. Features conversational memory, embedding-based retrieval, secure processing and low-latency inference without cloud …

Ollama, Local LLMs, Python, Embeddings, Conversational AI