Rajat Kumar is a results-oriented AI/ML and Data Science leader with 12+ years of experience building scalable, high-impact solutions across machine learning, deep learning, computer vision, NLP, and data engineering. He specializes in transforming complex business problems into intelligent, data-driven systems that deliver measurable outcomes. Currently, he leads AI initiatives at Biobrain Insights, where he works on advanced AI-powered platforms for survey automation, MLOps, and 3D spatial intelligence systems. He has also developed cutting-edge products like BioBrain, an AI-driven survey intelligence platform, and Twinn, a LiDAR-based real-time 3D reconstruction system. Previously, Rajat has held leadership roles across organizations such as Schlesinger Group, KPMG, Techomile, and Team Computers, where he built and deployed end-to-end ML systems for recommendation engines, fraud detection, contract intelligence, and predictive analytics at scale. His expertise spans Generative AI, LLMs, LangChain, LangGraph, Apache Spark, Kafka, Airflow, and cloud-native MLOps architectures. He has led cross-functional teams, architected enterprise-grade data platforms, and delivered AI solutions that significantly improved revenue, efficiency, and automation. Rajat holds an M.Tech in Data Science and Engineering from BITS Pilani and is currently pursuing advanced studies in Quantum Computing and Machine Learning at IIT Delhi. He is passionate about building intelligent systems that combine research-grade AI with real-world business impact.

Transforming the Future - From Computer Vision and Generative AI to Quantum Enhanced Intelligence Artificial Intelligence is rapidly shifting from isolated predictive models to integrated, large-scale systems capable of perception, generation, and decision support. With experience across traditional machine learning, computer vision, and Generative AI, my work has focused on building end-to-end AI systems that are not only accurate, but also scalable, robust, and deployable in real-world environments. In this talk, I will first discuss traditional machine learning from an applied systems perspective. This includes how supervised learning models are trained, how feature engineering still plays a critical role in structured data problems, and how model evaluation goes beyond accuracy to include metrics such as precision, recall, and ROC-AUC depending on the application. I will also highlight practical challenges in production, such as data drift, generalization, and maintaining model reliability over time. Next, I will move into computer vision, where deep learning has significantly advanced how machines interpret visual data. I will discuss convolutional neural networks and vision transformers, focusing on how models learn hierarchical representations of images. I will also share practical insights on handling real-world challenges such as limited labeled data, domain shift, and optimizing models for real-time inference in production systems using techniques like transfer learning and model compression. A major focus of the talk will be Generative AI, which is redefining modern AI systems. I will explain how transformer-based architectures and attention mechanisms enable large-scale models to generate text, images, and multimodal outputs. Beyond architecture, I will cover applied techniques such as prompt engineering, fine-tuning approaches (including parameter-efficient methods), and retrieval-augmented generation (RAG) for grounding outputs in external knowledge. I will also discuss challenges such as hallucinations, evaluation of generative outputs, and deploying LLMs efficiently at scale. Finally, I will connect these AI advancements to the emerging frontier of quantum computing. While still in early stages, quantum computing introduces new computational paradigms that could significantly impact AI—particularly in optimization, sampling, and high-dimensional search problems. I will discuss how quantum-enhanced algorithms may eventually complement classical AI pipelines, potentially improving training efficiency and enabling new forms of model optimization that are currently computationally expensive. This talk aims to provide a practical, experience-driven view of building modern AI systems today, while also exploring how quantum computing may shape the next generation of intelligent systems.