I’m a PreSales Manager and enterprise solutions architect who thrives on turning complex problem spaces into clean, scalable systems. Over 21+ years, I’ve focused on modernization, AI-enabled transformation, and building assets that make adoption practical and measurable.
Recently, I led validation and enablement for CAST Imaging’s MCP server rollout—testing multi‑IDE integrations, crafting reusable demo assets, and driving customer confidence with prescriptive playbooks and ROI narratives. I anchor solution design in clear business outcomes and responsible AI usage.
As a Generative AI Leader (Google Cloud certified), I bring provider‑agnostic, business‑level expertise across gen AI fundamentals, cross‑cloud patterns and offerings, techniques to improve model output quality, and strategies for responsible, outcome‑led adoption—applied across AWS and multi‑cloud environments, including NyaySamwad/NyayRAG.
I complemented this with NVIDIA Deep Learning Institute’s "Building RAG Agents with LLMs" training—centered on practical deployment and advanced orchestration. Key outcomes: composing LLM systems via internal and external reasoning; designing dialog management that maintains state and coerces information into structured formats; using embedding models for efficient similarity queries for retrieval and dialog guardrailing; and implementing, modularizing, and evaluating a RAG agent that answers questions on research papers without fine‑tuning. This strengthened my hands‑on practice in GPU‑accelerated workflows and performance‑aware deployment.
I founded and architected NyaySamwad, built atop NyayRAG—a proprietary retrieval and orchestration layer. It applies dialog state management, structured reasoning, and embeddings‑based guardrails to deliver predictable answers, and is deployed on AWS serverless with a cloud‑agnostic architecture for portability and performance. NVIDIA’s “Building RAG Agents with LLMs” sharpened the agent composition and evaluation approach, while the Generative AI Leader certification informs responsible adoption and business strategy across providers.
Beyond my professional work, I share learnings via my podcast and YouTube channel, and I enjoy experimenting in the kitchen or long‑distance running—both give me space to think through tough architecture problems.
Founded, architected and developed a scalable AI legal chatbot for Telegram and WhatsApp, evolving the core engine from a standard conversational model to a specialized Retrieval-Augmented Generation (RAG) system. This advanced solution is built to deliver highly accurate and context-aware legal information.
Designed and implemented a custom RAG solution named NyayRAG, trained on a curated corpus of public legal documents to provide domain-specific and authoritative responses.
Developed the system on AWS Bedrock, leveraging its foundation models and capabilities to power the RAG and conversational AI components.
Currently working on transitioning the system towards an Agentic RAG framework, creating multi-step workflows to automate and streamline responses for complex legal use cases.
Utilized AWS Lambda for business logic and Amazon DynamoDB for scalable chat data storage, ensuring high availability and cost-effective operations.
Impact: This project highlights deep expertise in business acumen, problem solving, solutions architecture, advanced generative AI development (RAG and Agentic systems), serverless computing, and the strategic use of AWS Bedrock to build sophisticated, domain-specific AI applications.