At Centre for Railway Information Systems (CRIS), I developed a predictive machine learning model to forecast train delays from historical transit data and improve estimated arrival times.
I also collaborated with cross-functional teams on data visualization tools, turning operational insights into reports for railway management.
At Outlier.ai, I evaluated LLM responses and authored ground-truth answers and prompts to support model training through RLHF. I documented edge cases, hallucinations, and biases in model outputs.
At CSIR-National Aerospace Laboratories (NAL), CEM Division, I’m developing and experimenting with diffusion and GAN-based generative models for computational engineering applications. My projects also explore RAG evaluation, secure SQL analysis, time-series forecasting, and India’s EV and grid feasibility.
