About Newsletter Talks Books Research Speaking AWS Blogs Subscribe
Principal GenAI SA · AWS · 20+ years
building the future, in real-time —

Mani
Khanuja.

Expert on scaling autonomous AI agents safely & efficiently, AI governance, and strategic leadership.

With 20+ years of experience, Mani builds Generative AI strategy for enterprise customers at AWS. Her current focus is scaling autonomous AI agents safely and efficiently — from building AI platforms from scratch to governing agentic systems at scale. Researcher, author, and speaker sharing insights in The Agentic Enterprise newsletter.

About me Read the newsletter
Agentic AI RAG LLMs Context Engineering Amazon Bedrock Distributed Training
Mani Khanuja
20+ Years in AI & Tech
About

Architecting the
future, with care.

Mani Khanuja is a Technical AI Leader and Principal Generative AI Solutions Architect at AWS with 20+ years of experience building AI platforms from scratch and driving enterprise AI strategy. She works directly with customers to build their Generative AI strategy, from architecture to production deployment at scale.

Her current focus is scaling autonomous AI agents safely and efficiently: developing stateful, memory-driven agents with personalization, advancing AI governance frameworks, and translating cutting-edge research into real-world enterprise systems. She is the author of Applied Machine Learning and High-Performance Computing on AWS and co-author of the upcoming The AI Steering Wheel.

A pioneering researcher, she co-authored the widely cited paper "Keyword search is all you need: Achieving RAG-level performance without vector databases using agentic tool use." She is also a recognized technical speaker at re:Invent, Grace Hopper Celebration, AI Engineer Summit, and AWS Summits worldwide. She resides in Seal Beach, California, where she stays active with long runs along the coast.

Connect on LinkedIn Read Newsletter
Mani Khanuja
20+
Years of experience
100+
Conferences
2
Books
12+
Papers reviewed
Gallery

On stage & in the field.

re:Invent, Grace Hopper, AWS Summits — the conferences where the work gets stress-tested.

Mani Khanuja on air at AWS re:Invent
AWS re:Invent · On Air
Mani Khanuja speaking at AWS Summit Toronto
AWS Summit Toronto — High-Performance Model Deployment
Mani Khanuja at Grace Hopper Celebration 2023
Grace Hopper Celebration · 2023
Books

Published works.

From high-performance computing to the agentic frontier — bridging theory and enterprise practice.

The AI Steering Wheel
+
Upcoming

The AI Steering Wheel

Unified Framework for Scaling Generative and Agentic Systems

Three interlocking layers — Strategy, Operations, and Engineering — keep AI products aligned from first decision through last deployment. Co-authored with Dr. Fouad Bousetouane.

Applied Machine Learning and High-Performance Computing on AWS
+
Published

Applied Machine Learning and High-Performance Computing on AWS

A comprehensive guide to building, training, and deploying ML models at scale on AWS. Co-authored with Farooq Sabir, Shreyas Subramanian & Trenton Potgieter.

YouTube

Featured talks & sessions.

Deep-dives on Agentic AI, RAG, Amazon Bedrock, and distributed training — top videos by views.

Keyword Search is All You Need: RAG-Level Performance Without Vector DBs
Top viewedRAG · Agentic
Building Context-Aware AI Agents with Amazon Bedrock AgentCore Memory
LiveBedrock
Personalization of AI Agents: Stateful Agents Using Memory at Scale
Top viewedAgentic AI
Enterprise Agentic AI in Practice: Patterns for Production Deployments
LiveEnterprise
Context Engineering for LLMs: Building Smarter AI Applications
Top viewedLLMs · Context
Distributed Training at Scale on AWS: HPC Deep Dive
Top viewedHPC
View Full Playlist on YouTube →
Newsletter

The Agentic Enterprise.

Insights on Agentic AI, ethics, enterprise deployment, and the future of autonomous systems — from The Agentic Enterprise newsletter.

LinkedIn · The Agentic Enterprise
Part 1: Who's the Boss? Navigating Ethics & Safety in Autonomous AI
As AI agents gain autonomy, who bears responsibility when things go wrong? This first installment explores the ethical frameworks and safety guardrails that must be designed into autonomous systems from the ground up — not bolted on after.
2026 Read
LinkedIn · The Agentic Enterprise
Part 2: The Battle of Agents: Why 86% of Organizations Are Losing Control
New data reveals that most enterprises deploying AI agents are struggling with governance and control. This issue breaks down why agent orchestration fails at scale and what patterns separate successful deployments from chaotic ones.
2026 Read
LinkedIn · The Agentic Enterprise
Part 3: Who Pays When Your Agent Hallucinates?
When an AI agent takes a wrong action based on a hallucinated fact — and real money or data is at stake — who is liable? This installment digs into accountability frameworks, insurance, and the legal landscape emerging around agentic AI.
2026 Read
LinkedIn · The Agentic Enterprise
Grand Theft Intelligence: Why Prompt Injection Is the New SQL Injection
Prompt injection is fast becoming the defining security vulnerability of the agentic era. Just as SQL injection reshaped database security two decades ago, prompt injection demands a complete rethinking of how we secure AI-powered applications.
2026 Read
All articles on Substack LinkedIn Newsletter
Research

Research & publications.

Peer-reviewed papers advancing retrieval-augmented generation, agentic systems, and applied machine learning.

Keyword Search is All You Need: Achieving RAG-Level Performance Without Vector Databases Using Agentic Tool Use
Mani Khanuja et al. · arXiv · arXiv:2602.23368
Demonstrates that agentic tool-use with traditional keyword search can match or exceed vector database retrieval in RAG pipelines, with significant cost and complexity advantages for enterprise deployments.
Read paper
The Amazon Nova Family of Models: Technical Report and Model Card
Amazon · arXiv · arXiv:2506.12103
Presents Amazon Nova, a new generation of state-of-the-art foundation models delivering frontier intelligence and industry-leading price performance, including Nova Pro (multimodal), Nova Lite, Nova Micro, Nova Canvas (image generation), and Nova Reel (video generation). Covers benchmarking, agentic performance, long context, and safety.
Read paper
Peer Reviewer · WACV 2024, IEEE/CVF Winter Conference on Applications of Computer Vision
Primary Reviewer · 10–12 Papers Reviewed
Served as a primary reviewer for IEEE/CVF WACV 2024, contributing expert evaluation to advance the computer vision research community.
Peer Review
Speaking

Stages & conferences.

Recognized technical speaker at the world's leading AI & cloud conferences.

AWS re:Invent
2022 — 2024
Grace Hopper
2023
AI Engineer Summit
2024
AWS Summit Toronto
2024
AWS Summit NYC
2024
AWS Summit SF
2023
QCon
2024
ODSC
2023
AWS Blogs

Selected AWS writing.

Technical deep-dives published on the AWS Machine Learning and Generative AI blogs.

AWS Machine Learning Blog · 2025
Amazon Bedrock AgentCore Memory: Building Context-Aware Agents
A deep-dive into Amazon Bedrock AgentCore Memory, how to build AI agents that maintain context across sessions, enabling genuinely personalized, stateful interactions that improve over time in enterprise deployments.
Read
AWS Machine Learning Blog · 2026
Build Agents to Learn from Experiences Using Amazon Bedrock AgentCore Episodic Memory
A technical walkthrough on building AI agents with episodic memory, enabling agents to retain context across sessions, learn from past interactions, and deliver personalized responses at enterprise scale using Amazon Bedrock.
Read
AWS Machine Learning Blog · 2025
Streamline GitHub Workflows with Generative AI Using Amazon Bedrock and MCP
Explore how the Model Context Protocol (MCP) connects Amazon Bedrock to GitHub, enabling AI agents to automate pull request reviews, issue triage, and code generation workflows, reducing developer toil and accelerating delivery.
Read
AWS Machine Learning Blog · 2025
Unlocking the Power of Model Context Protocol (MCP) on AWS
A comprehensive guide to MCP on AWS, how the Model Context Protocol enables AI agents to securely connect to data sources, tools, and APIs. Covers architecture patterns, security considerations, and practical implementation with Amazon Bedrock.
Read
AWS Machine Learning Blog · 2025
From Concept to Reality: Navigating the Journey of RAG from Proof of Concept to Production
Bridges the gap between RAG prototypes and production-grade deployments on AWS. Covers chunking strategies, retrieval optimization, evaluation frameworks, and the architectural decisions that determine success at scale.
Read
AWS Machine Learning Blog
Create a Next-Generation Chat Assistant with Amazon Bedrock, Connect, Lex, LangChain, and WhatsApp
Step-by-step guide to building a production-ready conversational AI assistant that combines Amazon Bedrock's foundation models with Amazon Connect and Lex for omni-channel customer engagement over WhatsApp.
Read
AWS Machine Learning Blog · Author archive
All AWS posts authored by Mani Khanuja
15+ deep-dive articles spanning Amazon Bedrock, SageMaker, distributed training, RAG, and agentic AI patterns. Browse the full author archive.
View archive
Stay close

Come build with me.

Three places. Same voice. Different cadences.

Substack · Weekly
manikhanuja.substack.com
Long-form deep dives on scalability, evaluation, security & governance.
Subscribe →
LinkedIn · Weekly
The Agentic Enterprise
A LinkedIn newsletter for practitioners shipping agents now.
Read on LinkedIn →
LinkedIn · Connect
Mani Khanuja
Reach out for speaking, collaboration, or to stay in the loop.
Connect →