Hari Hara Sudhan N

•Architected and implemented a production-ready Agentic AI platform leveraging Amazon Bedrock AgentCore Runtime, Strands Agents, and custom tool integrations for autonomous database querying.
•Engineered a multi-step agent workflow that performs schema discovery, SQL generation, query validation, execution, and business-friendly result synthesis.
•Designed serverless API-based access using AWS Lambda and Amazon API Gateway, enabling secure and scalable agent invocation.
•Implemented observability, logging, and runtime diagnostics through Amazon CloudWatch to support production monitoring and debugging.
•Delivered stakeholder demonstrations showcasing enterprise-grade agent orchestration, tool usage, and natural-language analytics capabilities.
•Developed a modular Multi-Agent system comprising Orchestrator, Design, Knowledge, Validation, and Artifact Agents to automate Architecture solution generation workflows.
•Implemented agent orchestration and tool-calling mechanisms for requirement analysis, knowledge retrieval, validation, and artifact creation.
•Leading the transition from proof-of-concept to production-ready platform through enhanced agent capabilities, workflow automation, and enterprise-grade controls.

• Built Retrieval Augmented Generation (RAG) application using LangChain FAISS, BedRock Anthropic & Cohere Text Embed model for GenAI Comprehensive Guide including details inside the Hyperlink.
• Implemented HR Policy RAG Application with Streamlit Chat UI for Internal Employee HR QA Purpose
• Built Model Validation GenAI App for Data Quality, Variable Description, EDA, Feature Creation, Model Design and Key Finding using AWS BedRock with Anthropic Claude model.
• Implemented MLOps Pipeline in SageMaker for Credit Risk model deployment.

• Optimized AI Extraction Layer for large size Fund Setup PDF Documents more than 200K Tokens that
extracts Data Points from using AWS Bedrock Anthropic Claude 4.5 Model.
• Reduced the execution time of Extraction for bigger file from 42 mins to 9.5 mins which is 77%
reduce in time of extraction using Concurrent OCR extraction for the large files.
• Dynamically adjusted the Max_Token limit to make AWS Bedrock perform faster without hitting the
Rate Limit Error, Throttling Exception and Token Limit error.
• Chased AWS Team to increase Token Limit from 200K tokens to 1M token for processing large
document with a single LLM Call.
• Collaborated with Client Tech Delivery Lead, Head of Data & Product Development as Single Point-of
Contact for overall AI Layer in Fund_Setup Project.

• Built Retrieval Augmented Generation (RAG) application using LangChain FAISS, BedRock Anthropic
& Cohere Text Embed model for GenAI Comprehensive Guide including details inside the Hyperlink.
• Implemented HR Policy RAG Application with Streamlit Chat UI for Internal Employee HR QA Purpose
• Built Model Validation GenAI App for Data Quality, Variable Description, EDA, Feature Creation, Model
Design and Key Finding using AWS BedRock with Anthropic Claude model.
• Implemented MLOps Pipeline in SageMaker for Credit Risk model deployment.

• Built PDF Contract Document Legal Gap Analysis using AWS BedRock, Anthropic Claude Sonnet, AWS Textract, PyPDF and Frontend UI using Streamlit and Deployed in Prod Server with One-Click Upload GenAI LLM Analysis.
• Deployed Yield Prediction model using Sagemaker pipeline.
• Build the Entire Infrastructure for Database migration in CloudFormation IaaC with services including EC2, Elastic Load Balancer, Aurora DB including EC2 User Data.

• Built PDF Contract Document Legal Gap Analysis using AWS BedRock, Anthropic Claude Sonnet, AWS
Textract, PyPDF and Frontend UI using Streamlit and Deployed in Prod Server with One-Click Upload
GenAI LLM Analysis.
• Deployed Yield Prediction machine learning model using Sagemaker pipeline in AWS.
• Build the Entire Infrastructure for Database migration in CloudFormation IaaC with services
including EC2, Elastic Load Balancer, Aurora DB including EC2 User Data.

• Built a Data Pipeline for Analytics Dashboard using AWS Lambda & CDK in Python.
• Used Flask for Building API for Business Hotel Booking and Flight booking product for Team Events feature.

• Proof-of-Concept using HuggingFace LLM Models BERT & Llama for interaction with Txt/Word
Document.
• Built a Data Pipeline for Analytics Dashboard using AWS Lambda & CDK in Python.
• Used Flask for Building API for Business Hotel Booking and Flight booking product for Team Events
feature.

• Performed Text Analysis/ NLP on Social Media data to improve customer experience using Flask Web Application.
• Migrated PL/pgSQL Stored Procedure of MySQL to DBT Framework for Redshift. Built data pipeline for Ecommerce client using DBT, Airflow, AWS & RedShift.

• Performed Text Analysis/ NLP on Social Media data to improve customer experience using Flask Web
Application.
• Migrated PL/pgSQL Stored Procedure of MySQL to DBT Framework for Redshift. Built data pipeline
for Ecommerce client using DBT, Airflow, AWS & RedShift.

• Performed EDA on Flagship Project - derived insights using Python (Numpy, Pandas, Matplotlib).
• Working Closely with Product Managers and Engineering Managers to analyze the Resource Allocation Capacity and finalize the project priority based on the Customer Urgency.
• Migrated the entire allocation datastore to Service now and performed end to end project allocation to visualization in Servicenow.

• Data creation, curation, annotation and analysis to help develop, test and train the Alexa AI.
• Improved the Alexa Shopping Goal Success Rate (GSR) from 65% to 92% by analysis purchase patterns of US Voice Shopping Market place.

• Data creation, curation, annotation and analysis to help develop, test and train the Alexa AI.
• Improved the Alexa Shopping Goal Success Rate (GSR) from 65% to 92% by analysis purchase
patterns of US Voice Shopping Market place.