Archish Rai Kapil

I came to data science by an unconventional route. After studying History, I taught myself statistics, programming, and machine learning, eventually building a career that has now spanned more than a decade across aviation, consulting, and healthcare.

My work has taken me through a range of data science and AI problems, and more recently into Responsible AI, where I focus on evaluating how large language models behave in enterprise settings. Along the way I've authored research papers, designed Responsible AI evaluation frameworks, developed enterprise AI libraries, and continued to document that journey through DataVedas.

DataVedas began in 2018 as a book of more than 630 pages, bringing together everything I had learned while navigating data science through books, research papers, documentation, and experimentation. Over time it grew into something much larger, a structured platform dedicated to helping others learn the subject with greater clarity.

The name draws inspiration from Sage Vyasa, who is traditionally credited with compiling the Vedas into an organised body of knowledge. In much the same spirit, DataVedas brings together concepts from across data science into a single, structured resource.

Today it continues to expand across data science, machine learning, MLOps, generative AI, Responsible AI, and practical implementation, with the same objective it started with: making complex subjects easier to navigate without losing their depth.

My current work focuses on Responsible AI and the evaluation of large language models. I design enterprise frameworks that assess how AI systems behave, measuring qualities such as factual accuracy, fairness, bias, robustness, safety, and task-specific performance. The work spans both research and implementation, developing evaluation methodologies and production-ready libraries that help organisations deploy AI systems more responsibly.

One question continues to drive much of this work: how do we evaluate a model when there is no single correct answer? As generative AI becomes part of real-world decision-making, answering that question is becoming increasingly important.

Data science is often taught as disconnected topics. My goal with DataVedas is to bring those ideas together into a structured learning path that emphasizes understanding alongside practical application.

Data science is constantly evolving, but understanding the fundamentals never goes out of style. DataVedas continues to grow with that in mind, bringing together theory, practical implementation, and Responsible AI into a single, structured resource for anyone who enjoys learning.