Software Intern – NOOTUS Innovations LLP (TaxGPT / LLM Playground & Model Registry)
Built and evaluated an LLM-driven product (TaxGPT/LLM Playground) enabling real-time comparison of 10+ large language models. Focused on configuring prompts and running evaluation workflows to assess model outputs during development cycles. Used this pipeline as part of an LLM evaluation loop for the application’s chat and registry features. • Real-time LLM model comparison across LLaMA, GPT, and Gemini variants • Prompt-to-chat panel interactions managed via state workflows (system prompts, chat panels, configuration) • LLM Playground and Model Registry module development for experimentation • Integrated secure RAG file-upload flows to support evaluation contexts
2025 - 2025