Ramkumar A
Bengaluru
Summary
Generative AI Architect with 10+ years of technical experience and 4+ years of extensive hands-on expertise in GenAI, Multi-Agent systems, LLM engineering, and cloud-scale AI architecture across Azure, AWS, and GCP.
Overview
10
10
years of professional experience
2
2
Certifications
Work History
AI/ML Specialist → GenAI Architect
Accenture
Bangalore
05.2019 - Current
- Experience on ML libraries like PyTorch, Tensorflow.
- Proven track record delivering 12 AI projects, including 5 production deployments involving RAG, multi-agent orchestration, vector search, model fine-tuning, Agentic AI bot and enterprise LLMOps
- Designed multi-agent workflows, LLM fine-tuning pipelines, and vector-based retrieval architectures.
- Worked on RAG-based chatbot and implemented solutions in Vector databases (Kendra, Pinecone)
- Expert in designing scalable, secure AI architectures, optimizing AI inference cost/performance, and building Modular Agentic Systems using frameworks such as LangGraph, CrewAI, Bedrock-agent-core, Strands and LangChain.
- Designed structured and goal-driven prompts that enable LLMs to reason, use tools, and produce reliable, machine-readable outputs for automation
- Delivered multiple production GenAI deployments, optimizing compute cost using quantization, caching, and scalable inference strategies.
Assistant Manager
Smartlytics Innovations Pvt Ltd
Bengaluru
01.2018 - 05.2019
· Involved completely in implementing Database and Sisense set up independently.
· Involved in Data harmonization & visualization, Data Harmonization ETL (Pentaho Kettle), Data Uploads in Live.
· Have used ETL tools to extract data from different data sources.
Analyst Consultant
Kantar Analytics – Gain Theory, Bangalore
Bengaluru
07.2015 - 01.2018
· Involved in Data harmonization & visualization, Data Harmonization ETL (SAS), Data Uploads in Live.
· Also Involved in Data Visualization/Dashboard Development using SISENSE i.e. The only business analytics software that lets you easily prepare and analyze big, scattered datasets.
· Used Tableau and Power BI to build dashboards.
Education
Bachelor of Engineering - EEE
M. Kumarasamy College of Engineering
Karur, Tamil Nadu11-2014
Skills
- Generative AI Architecture (Bedrock Models, OpenAI, GPT-4o / O1 / Claude / Gemini / Llama 32 / )
- Multi-Agent System Design ( Bedrock-Agent-Core, Strands, CrewAI, LangGraph)
- LLM Frameworks: LangChain, Semantic Kernel, LlamaIndex
- Model Fine-Tuning & Data Strategy: Data collection, cleaning, preprocessing, synthetic data pipelines
- Cloud Platforms: AWS Bedrock, GCP Vertex AI,Microsoft Azure
- LLMOps & DevOps: Azure DevOps, CI/CD, model monitoring, prompt evaluation, governance
- Data Engineering: Apache Spark, ETL Pipelines, Data Lakes, Real-time event-driven systems
- Vector Databases: Kendra, Search and Conversation, Pinecone, Azure AI Search
- Infrastructure: Kubernetes, Docker, Serverless Architectures
Certification
AWS Machine Learning Specialist
Affiliations
Project 1: RAG-based AI Chatbot (productionized) (Tested on all 3 cloud platforms)
Description: Developed a chatbot using (RAG) architecture to deliver contextual responses to user queries. This chatbot is developed in AWS, Azure and GCP and AWS version is productionized. Tools Used: AWS Lambda, Amazon Bedrock, Amazon Kendra, S3, Python
Roles & Responsibilities:
Design and implement a serverless chatbot using AWS Lambda integrated with Amazon Bedrock-hosted models.
Utilize Amazon Kendra as the vector database for intelligent document retrieval in the RAG pipeline.
Build modular Lambda functions to manage query parsing, document retrieval, and generative response generation.
Perform prompt engineering to ensure accurate, context-aware responses aligned with user expectations.
Project 2: AI-based Sales Representative Performance Evaluation System (Integrated with Blue Prism)
Description: Developed an AI-driven evaluation system to assess the communication effectiveness and sales skills of medical representatives during interactions with doctors. The system analyzes meeting transcripts using prompt-based evaluation criteria and generates detailed performance reports. It supports multilingual evaluation — prompts are in English, while reports can be generated in custom languages. Integrated seamlessly with Blue Prism for ETL automation and AWS Bedrock for AI processing.
Tools Used: Blue Prism, AWS Bedrock, S3, Python, NLP, Prompt Engineering
Roles & Responsibilities:
Designed an AI workflow to evaluate medical sales conversations based on predefined performance criteria.
Integrated Blue Prism to handle ETL processes and automate transcript ingestion.
Leveraged AWS Bedrock for LLM-based transcript analysis and performance scoring.
Implemented multilingual reporting capability to generate results in user-defined languages.
Optimized prompt instructions for consistent, objective evaluation across diverse transcripts.
Description: Developed a chatbot using (RAG) architecture to deliver contextual responses to user queries. This chatbot is developed in AWS, Azure and GCP and AWS version is productionized. Tools Used: AWS Lambda, Amazon Bedrock, Amazon Kendra, S3, Python
Roles & Responsibilities:
Design and implement a serverless chatbot using AWS Lambda integrated with Amazon Bedrock-hosted models.
Utilize Amazon Kendra as the vector database for intelligent document retrieval in the RAG pipeline.
Build modular Lambda functions to manage query parsing, document retrieval, and generative response generation.
Perform prompt engineering to ensure accurate, context-aware responses aligned with user expectations.
Project 2: AI-based Sales Representative Performance Evaluation System (Integrated with Blue Prism)
Description: Developed an AI-driven evaluation system to assess the communication effectiveness and sales skills of medical representatives during interactions with doctors. The system analyzes meeting transcripts using prompt-based evaluation criteria and generates detailed performance reports. It supports multilingual evaluation — prompts are in English, while reports can be generated in custom languages. Integrated seamlessly with Blue Prism for ETL automation and AWS Bedrock for AI processing.
Tools Used: Blue Prism, AWS Bedrock, S3, Python, NLP, Prompt Engineering
Roles & Responsibilities:
Designed an AI workflow to evaluate medical sales conversations based on predefined performance criteria.
Integrated Blue Prism to handle ETL processes and automate transcript ingestion.
Leveraged AWS Bedrock for LLM-based transcript analysis and performance scoring.
Implemented multilingual reporting capability to generate results in user-defined languages.
Optimized prompt instructions for consistent, objective evaluation across diverse transcripts.
Timeline
AI/ML Specialist → GenAI Architect
Accenture
05.2019 - Current
Assistant Manager
Smartlytics Innovations Pvt Ltd
01.2018 - 05.2019
Analyst Consultant
Kantar Analytics – Gain Theory, Bangalore
07.2015 - 01.2018
Bachelor of Engineering - EEE
M. Kumarasamy College of Engineering