Production systems, research prototypes, and shipped products. Each one solving a concrete problem.
Intelligent content retrieval and generation across 150+ eBooks using OCR, RAG, and quantized LLMs.
Section-aware resume analysis using Llama 2, RAG, and OCR. Built on Django for end-to-end HR workflows.
German-to-English audio translation with targeted information extraction using Whisper, Hugging Face, and RAG.
A responsive technical-support chatbot powered by GPT-3 with intent detection and context retention.
Visual product search and personalized recommendations using TensorFlow, Keras, Flask, and PostgreSQL.
CNN + LSTM model trained on COCO dataset to describe images in English with transfer learning from Inception.
Daily attendance system using CNN-based face recognition with augmentation, segmentation, and a Flask UI.
Audio-based genre classification using Mel spectrogram, MFCC features, and GRU+LSTM models.
Encoder-Decoder LSTM model trained on a 25K-sentence Hindi translation corpus.
Bidirectional LSTM trained on 6,500+ medium articles using N-gram preparation for next-word prediction.
Compared 6 algorithms (SVR, RFR, KNN, LSTM, GRU, LSTM+GRU) for 10-day stock prediction. GRU won.
EDA on 10K+ apps to surface 20+ insights about ratings, installs, categories, and pricing trends.
Tableau dashboard with 10+ insights on recovery, mortality, and forecasting case growth.