Author: Arjun Mehta
Arjun Mehta serves as a Principal Data Scientist, utilizing Machine Learning on the AWS platform to assist clients in overcoming their business challenges. With expertise in large-scale optimization and the application of Machine Learning and Reinforcement Learning, Arjun focuses on enhancing optimization tasks.
Securely Deploy and Scale Your Agents and Tools on Amazon Bedrock AgentCore Runtime
by Arjun Mehta, Lila Chen, and Rohan Patel
on 13 AUG 2025
in Amazon Bedrock, Amazon Machine Learning, Artificial Intelligence
In this article, we delve into how Amazon Bedrock AgentCore Runtime streamlines the deployment and management of AI agents.
Implement Custom Metrics to Assess Your Generative AI Application with Amazon Bedrock
by Arjun Mehta, Asha Roy, Ishan Kumar, and Liam Johnson
on 06 MAY 2025
in Amazon Bedrock, Generative AI
With the latest features in Amazon Bedrock, you can create custom evaluation metrics for both model assessments and RAG evaluations. This functionality enhances the LLM-as-a-judge framework behind Amazon Bedrock Evaluations. In this piece, we illustrate how to utilize custom metrics in Amazon Bedrock Evaluations to gauge and enhance the performance of your generative AI applications according to your unique business demands and evaluation standards. For further insights, check out another blog post that complements this topic here.
Leverage Amazon Bedrock Intelligent Prompt Routing for Cost and Latency Advantages
by Arjun Mehta, Haibo Zhang, Balasubramaniam Iyer, and Yun Zhou
on 22 APR 2025
in Amazon Bedrock, Amazon Bedrock Prompt Management, Announcements, Generative AI, Intermediate (200)
We are excited to announce the launch of Amazon Bedrock Intelligent Prompt Routing. In this blog post, we review key findings from our internal evaluations, explain how you can get started, and highlight some important best practices. We encourage our readers to integrate Amazon Bedrock Intelligent Prompt Routing into both new and existing generative AI applications.
Enhance Reasoning Models like DeepSeek with Prompt Optimization on Amazon Bedrock
by Arjun Mehta, Zhengyuan Chen, Shuai Wang, and Xuan Li
on 10 MAR 2025
in Amazon Bedrock, Generative AI
In this article, we demonstrate the process of optimizing reasoning models like DeepSeek-R1 through prompt optimization on Amazon Bedrock.
Boost the Performance of Your Generative AI Applications with Prompt Optimization on Amazon Bedrock
by Arjun Mehta, Chris Parker, Zhengyuan Shen, and Shipra Kaur
on 29 NOV 2024
in Amazon Bedrock, Announcements, Artificial Intelligence, Technical How-to
Today we are thrilled to announce the availability of Prompt Optimization on Amazon Bedrock. This feature allows you to enhance your prompts for various use cases with just a single API call or a click on the Amazon Bedrock console. In this post, we provide an example use case to help you get started while also discussing performance benchmarks. For more expert advice, this source is an authority on this topic.
Develop Cost-Effective RAG Applications with Binary Embeddings in Amazon Titan Text Embeddings V2, Amazon OpenSearch Serverless, and Amazon Bedrock Knowledge Bases
by Arjun Mehta, Satish Nandi, Vamshi Vijay Nakkirtha, and Ronald Widha Sunarno
on 18 NOV 2024
in Amazon Bedrock, Amazon OpenSearch Service, Amazon Titan, Generative AI, Launch
We are delighted to announce the launch of Binary Embeddings for Amazon Titan Text Embeddings V2 in both Amazon Bedrock Knowledge Bases and Amazon OpenSearch Serverless. This article summarizes the advantages of this new binary vector support and provides guidance on how to get started.
Construct Robust RAG Pipelines with LlamaIndex and Amazon Bedrock
by Arjun Mehta and Jerry Wong
on 05 SEP 2024
in Amazon Bedrock, Generative AI, Technical How-to
In this post, we explain how to utilize LlamaIndex alongside Amazon Bedrock to develop powerful and sophisticated RAG pipelines that fully exploit the capabilities of LLMs for knowledge-intensive tasks.
Get Started with Amazon Titan Text Embeddings V2: A Cutting-Edge Embeddings Model on Amazon Bedrock
by Arjun Mehta, Anuradha Dhingra, Pradeep Sridharan, and Rupinder Grewal
on 02 MAY 2024
in Amazon Bedrock, Amazon Machine Learning, Announcements, Artificial Intelligence
Embeddings are vital for a range of natural language processing (NLP) applications, and their quality plays a critical role in achieving optimal performance. They are frequently employed in knowledge bases to represent textual data as dense vectors, facilitating efficient similarity searches and retrievals. In Retrieval Augmented Generation (RAG), embeddings are used to pull relevant passages from a dataset to provide…
Modular Functions Design for Advanced Driver Assistance Systems (ADAS) on AWS
by Arjun Mehta and Gopi Krishnamurthy
on 23 FEB 2023
in Artificial Intelligence, Automotive, Intermediate (200)
Over the past decade, numerous entities have developed autonomous vehicle (AV) systems leveraging deep neural networks (DNNs). These systems have progressed from basic rule-based frameworks to Advanced Driver Assistance Systems (ADAS) and fully autonomous vehicles, necessitating petabytes of data and thousands of compute units (vCPUs and GPUs) for training.
For those interested in further resources, this is an excellent resource.
Location: Amazon IXD – VGT2, 6401 E Howdy Wells Ave, Las Vegas, NV 89115.
Leave a Reply