At Ethara AI, I trained and fine-tuned LLMs using curated datasets, then validated responses with Multimango to identify errors and refine training data and prompts.
At i2V Systems, I trained and fine-tuned deep learning models for real-time object detection and image classification. I built CNN pipelines in Python, PyTorch, and OpenCV, and benchmarked models using precision, recall, F1-score, and mAP.
On my Zomato analysis project, I built 10 interactive Tableau dashboards to explore sales across cities, cuisines, and restaurants, and forecasted revenue. I also analyzed city-wise ratings and top-performing restaurants.
I built a content-based movie recommendation system using TF-IDF and cosine similarity, integrating TMDB data and a Streamlit interface. My projects also include a wheat disease classifier that I benchmarked across five CNN architectures, and I hold AWS, Microsoft Azure AI-900, and MongoDB certifications.
