Abhishek Bhamidipati

About Abhishek Bhamidipati

I’m a Generative AI Engineer and a CMU graduate who designs production-scale AI systems that automate enterprise workflows, enhance decision-making and save millions in operational costs.In my current role, I built Marketing AI Studio, a multi-agent system leveraging Google ADK, LangChain, and FastAPI to autonomously generate marketing content across formats, saving $1.7M annually by eliminating third-party tools. I also developed Rep Nudges (Gemini 2.0 Flash), Scribe AI (Salesforce-integrated outreach generator using Gemini Pro + Milvus), and Sales Infobot (Dialogflow CX), collectively improving productivity, automation and user trust across Verizon’s sales ecosystem.Beyond enterprise work, I’ve worked on multiple independent GenAI projects including Lecture Loom, an interactive RAG-based learning platform for Vanderbilt University using Whisper, Pinecone, and GPT-4, and AdPulse, a fine-tuned Bloom 1b7 model for ad generation via PEFT on Hugging Face.My technical toolkit spans LLM models, LangChain, Crew AI, LangGraph, Hugging Face, FastAPI, Docker, Apigee, vector databases (Milvus, Pinecone, Chroma), RAG pipelines, and MLOps for scalable deployment.I’m driven by a single goal to transform how businesses and users interact with information through intelligent, autonomous AI systems. I am passionate about learning new things and solving problems. Nothing stokes my fire more than pouring my all into a project and watching it flourish. I have got passion, a keen eye for detail and a curious mind that work together to achieve some amazing things.Open to: Generative AI Engineering, Applied LLM Engineer, or Machine Learning Engineer roles focused on production AI, multi-agent systems, and enterprise automation.

Experience

1. Generative Ai Software Engineer

   Innova Solutions

   Sep 2024 — Present

   Currently working on \"TYS (Transfer Your Service)\", a multi-agent system using Gemini 2.5 Flash, to autonomously resolve service transfer failures across Consumer and Business accounts. Built using Google ADK, FastAPI and MCP server integration to orchestrate AI agents along with Elasticsearch-powered RAG for grounded retrieval from enterprise knowledge bases, thereby reducing manual intervention by Verizon sales reps and improving transfer success rates- Architected \"Marketing AI Studio\", a multi-agent autonomous content generation platform using LangGraph, LangChain and FastAPI, generating emails, blogs and social media content at scale and delivering $1.7M annual cost savings by reducing third-party dependencies. Engineered persistent agent memory with PostgreSQL checkpointers and implemented observability using Galileo telemetry- Engineered \"Rep Nudges,\" an intelligent FastAPI and Gemini 2.0 Flash-powered system that transforms complex sales metrics into prioritized recommendations for Verizon\'s sales representatives, reducing decision-making time and improving sales productivity through intelligent task prioritization- Built “Scribe AI,” a FastAPI-based application integrated with Salesforce CRM, using Gemini Pro 2.0 and Milvus vector DB for real-time retrieval of Verizon plans/promotions, enabling sales reps to generate high-quality outreach emails and reducing writing time by 78%- Developed “Sales Infobot” using Dialogflow CX and Google Cloud Datastore, delivering instant product, service and policy information with 92% intent accuracy and 65% faster query resolution. Collaborated with end-users to resolve issues, reducing negative reviews by 55% and significantly improving trust- Created an automated testing framework using Gemini to generate QA pairs from Verizon’s knowledge base, with evaluation pipelines measuring relevancy, groundedness, and contextual accuracy, reducing manual testing effort by 85%.

Education

• Carnegie Mellon University

  Master of Science - MS, Electrical and Computer Engineering

• Vellore Institute of Technology

  B. Tech, Electronics and Communications Engineering