I’m Vivek Bose, currently studying at ISI Kolkata (Indian Statistical Institute).
I’m driven by curiosity and experimentation — I actively explore the frontiers of the AI field: LLMs, agentic systems, RAG pipelines, and beyond. Rather than staying with off-the-shelf models, I train and fine-tune transformers myself, and I’m building up a collection of models on Hugging Face — including LoRA adapters, fine-tuned text classifiers, embedding/retrieval models, and vision models.
What I build
- Agentic AI — assistants that analyze live market data, reason over news with source citations, and answer questions in natural language.
- RAG systems — domain Q&A with document upload and section-level citations across multiple LLM providers.
- Real-time forecasting — dashboards for equities, indices, and cryptocurrencies with deployment-ready models.
- Trained models — transformers fine-tuned for finance, banking, tax retrieval, product retrieval, and medical imaging (see my Hugging Face).
Connect
- Portfolio: vivekbose.com/portfolio
- GitHub: github.com/bravo2024
- Hugging Face: huggingface.co/vivekkopthsd