Machine Learning to Gen AI Agents

Description

This book began as notes to myself.
Three years ago, I started learning machine learning at my kitchen table in Calgary after my family had gone to sleep. I had spent over fifteen years building enterprise systems at Accenture in SAP security and compliance. AI was reshaping everything, and I wanted to understand it properly — not just the buzzwords. So I started writing things down in my own words: the analogy that clicked, the example that made an abstract idea concrete, the story I could explain to a colleague over coffee.
Those notes grew. Through the Advanced Programme in AI for Leaders at IIM Calcutta — live classes at three in the morning from Calgary — and through building a pharmaceutical research AI copilot in production, every lesson went into those notes. One evening my wife saw the stack. “Why don’t you turn this into a book?” I looked at the notes and realized they already were one.
A note on currency. I once spent weeks studying sigmoid activation functions — only to find the field had moved on to SwiGLU and ReLU variants. Tutorials were broken, libraries deprecated, APIs changed. This book covers only what is current and relevant today. Older concepts appear only to build intuition — never as a recommendation. Your time is too valuable to spend on what the field has already left behind.
What this book is. Concepts first, always. No code. No equations. Stories and analogies that make machine learning intuitive — for leaders making AI decisions, architects designing systems, and practitioners who want the full picture without a mathematics background. Every framework, tool, and technique in these eighteen chapters was deliberately chosen for relevance today — obsolete models, deprecated libraries, and superseded approaches were left out entirely. One continuous journey from foundations to the frontier. No dead ends.
It took three years of learning, building, and writing to produce these chapters. I hope they save you some of that time.

About The Author

Prashant is a Security Delivery Manager, AI & ML Architect, and SAP Generative AI Developer at Accenture, where he has spent more than fifteen years at the intersection of enterprise technology, security, and artificial intelligence. His work spans SAP security and compliance, AI-driven automation, and the design of intelligent systems for global enterprise clients.
A graduate of the Advanced Programme in AI for Leaders at the Indian Institute of Management Calcutta — one of India’s most prestigious management institutions — Prashant completed a rigorous year of study that included attending live sessions at three in the morning from Calgary, Canada. That commitment to learning, even across time zones and at personal cost, reflects the same spirit that gave rise to this book.
He holds SAP AI Core certification and has applied machine learning and generative AI across domains ranging from pharmaceutical research to enterprise security. Among his independent projects is a production-grade AI copilot for pharmaceutical research — combining a knowledge graph, vector databases, molecular structure generation, and multi-agent workflows to accelerate compound identification and drug discovery research.
Beyond his enterprise work, Prashant is actively building at the frontier of applied AI. He is currently developing a FinTech intelligence system that combines social media signals with historical market data to surface predictive insights for share market analysis — an exploration of how real-time narrative and quantitative history can be made to speak to each other. He is also working on a Privacy Gateway framework for safe agentic AI — a novel architecture combining taint-aware data flows, agent control planes, and enterprise-grade separation of duties to make autonomous AI systems trustworthy by design. Both projects are works in progress, reflecting a conviction that the most interesting problems in AI today are not yet solved.