About Me
I'm Prashant Dudami, an AI/ML Architect with over 15 years of experience building enterprise-scale systems. My work focuses on making AI practical, cost-effective, and reliable for real-world business applications.
My Journey
My career has been a progression from data architecture to AI systems. At CSG International, I developed a patented methodology for data lake management that would later prove foundational for AI applications. This work earned a U.S. patent and has since been cited by R&D teams at Intuit, Johnson & Johnson, and Mastercard.
At Amazon Web Services, I worked as a Cloud Application Architect, leading technical discussions with enterprise clients and shaping AWS product strategy. This experience gave me deep insight into how organizations adopt AI technologies and the barriers they face.
Today, I focus on production Generative AI systems, particularly RAG (Retrieval-Augmented Generation) implementations that combine the power of LLMs with structured enterprise data. I've developed methodologies like TOON (Token Optimized Object Notation) that help organizations reduce their AI operational costs by 40-60%.
I believe in sharing knowledge openly. My open-source contributions and technical publications aim to help the broader technology community build better AI systems.
Professional Experience
Lead Staff Software Engineer
Design authority for production Generative AI systems. Implementing RAG systems integrating structured hospitality data with LLM-powered insights. Architecting hallucination-resistant chatbots leveraging real-time page context.
Solutions Architect
Collaborated with enterprise clients to define requirements and translate them into scalable application architectures. Researched and implemented AI services on Microsoft Azure.
Cloud Application Architect
Technical leader working alongside customer business and infrastructure teams. Led design discussions with IT executives and influenced AWS product strategy by representing the voice for developers internally.
Enterprise Solutions Architect
Redesigned eCommerce platform architecture using micro-service patterns. Developed automated data pipelines for AI-driven reporting and automation use cases.
Technical Lead
Led design and development of Ascendon, a large-scale cloud-based SaaS platform. Built scalable microservices for a DataLake architecture on AWS. Granted U.S. Patent for data lake methodology.
Technical Expertise
Key Achievements
Granted U.S. Patent No. 10,846,307 B1 for 'System and Method for Managing Interactions with Data Lake' - cited by Fortune 500 companies including Intuit, Johnson & Johnson, and Mastercard.
Creator of TOON Converter (Token Optimized Object Notation) - a Python library that reduces LLM token consumption by 40-60%, published on PyPI and GitHub.
Published technical articles on enterprise AI, LLM optimization, and data architecture reaching thousands of practitioners in the AI/ML community.
Let's Work Together
Whether you need help with LLM infrastructure, RAG implementation, or AI cost optimization, I'm here to help.