Headline

Full-Stack Agentic AI Engineer building RAG pipelines, AI agents & AI workflows

About

I am a Full-Stack Agentic AI Engineer and with 4.2+ years of experience building systems where traditional full-stack stability meets Generative AI. My expertise lies in bridging the gap between robust backend engineering and production-grade AI orchestration. The GenAI Edge I specialize in building autonomous AI agents and RAG (Retrieval-Augmented Generation) systems that go beyond basic chat. Using LangChain and LangGraph, I design context-aware workflows that leverage semantic retrieval and vector databases like Pinecone and ChromaDB. At Infosys, I’ve delivered high-impact solutions including: AI-Driven Code Intelligence: Automated code review and refactoring systems. Media Intelligence: Pipelines for extracting and processing insights from complex media. Agentic Integrations: Building MCP-style (Model Context Protocol) integrations and tool-based AI architectures. Core Engineering Foundation While I focus on AI, my foundation is built on scalable, enterprise-grade development. I specialize in the MEAN stack (MongoDB, Express, Angular, Node.js) and Python/FastAPI to build microservices that handle real-time data via Socket.IO. I am comfortable managing the entire lifecycle of a product—from designing responsive Angular frontends to containerizing services with Docker and deploying to cloud environments like AWS and Azure. Cloud & LLM Proficiency I have deep experience integrating and optimizing models across major platforms: AWS Bedrock (Anthropic, Titan, Llama) Azure OpenAI Services Google Vertex AI I am passionate about building the next generation of AI-native products—specifically focus-driven AI agents and RAG systems that solve real-world business bottlenecks. Technical Snapshot Languages: TypeScript, JavaScript, Python, HTML/CSS. Frameworks: Angular, Node.js, Express.js, FastAPI. AI/ML: LangChain, LangGraph, Vector Search (Pinecone/Chroma), Prompt Engineering, Semantic Retrieval, LLM Observability. Databases: MongoDB, PostgreSQL, Vector DBs. Cloud & DevOps: AWS (EC2, ECS, Bedrock), Azure OpenAI, Vertex AI, Docker, Git, CI/CD.

Location

Pune District

AI stack

antigravity

claude code

codex

github copilot

langgraph

Work experience

Senior Software Engineer AI at NICE Actimize, Aug 2026 - Present · 2 mos, Pune City

Senior Analyst – AI/ML Computational Science at Accenture, Feb 2026 - Aug 2026 · 7 mos, Pune District, • Building a conversational procurement assistant for an Oil & Gas enterprise on Microsoft Teams and web, where business users query enriched vendor catalogs, explore PO history, and receive structured vendor recommendations in natural language. This eliminated their dependency on SAP access and manual analyst queries. • Built the catalog enrichment agent on AWS Strands Agents with a typed tool-calling interface mapping each tool to pre-computed data sources. Built the solution on Amazon Bedrock AgentCore with session memory, Bedrock Guardrails for output safety, and role-based tool access, ensuring every response stays grounded and citable. • Implemented the RAG retrieval pipeline using hybrid kNN+BM25 search in Amazon OpenSearch, and integrated multiple LLMs– Claude Sonnet and Amazon Nova Lite– with task-complexity-based routing to balance response quality, latency, and cost across the assistant’s query types. • contributed to the data enrichment pipeline with LLM-assisted PO normalization, country-to-region mapping, and derived enrichment columns, feeding automated OpenSearch bulk ingestion and weekly Bedrock Knowledge Base refreshes. This removed manual data curation from the catalog update cycle. • Built a FastAPI service layer exposing enriched catalog data to downstream consumers including the SAP Joule search agent, unifying catalog access across internal agents and direct query consumers through a single API.

Specialist Programmer L2 at Infosys, Oct 2024 - Feb 2026 · 1 yr 5 mos, • Led development of a GenAI-powered code review and auto-refactor module for Infosys AI Toolchain as the primary developer. Shipped it as both a standalone Python package for pipeline integration and a FastAPI service with WebSocket endpoints for autonomous Git-triggered review, reducing manual code review effort across teams using the platform. • Built a multi-step LangGraph agentic workflow where a review node scans code against modular rule sets, a fix node applies LLM-generated corrections, and a verify node confirms validity. On failure, conditional edges loop back automatically, so the system self-corrects without manual intervention. • Integrated ChromaDB to persist project context and prior review state across sessions, enabling consistent cross-file fix decisions on large multi-module codebases and avoiding contradictory fixes between related files. • Generated self-contained interactive HTML reports per review run with collapsible findings, category filters, and full-text search, giving teams a single-click audit trail without needing separate reporting tools. • Designed an experimental MCP (Model Context Protocol) server wrapping core review utilities, allowing external AI coding assistants to programmatically trigger reviews and consume structured results.

Digital Specialist Engineer at Infosys, Jul 2022 - Oct 2024 · 2 yrs 4 mos, Pune/Pimpri-Chinchwad Area, • Introduced a GenAI-driven content generation layer to the Personalized Smart Video (PSV) platform. This replaced a multi-stage manual studio workflow with single-prompt video draft generation, cutting content creation turnaround from days to minutes. • Built the LLM orchestration pipeline that ingests user-provided documents and web URLs, extracts and summarizes content, and generates structured scene-level scripts with VideoJS-compatible metadata. Users could go from raw documents to a structured video draft without manual scripting. • Implemented LLM-based auto-tagging and semantic description generation for image and video assets, extracting frame-level content descriptions and storing them alongside system metadata to power relevance-based media selection. • Built a semantic media retrieval layer using LangChain and Pinecone, replacing rule-based asset selection with embedding-based search over the tagged media library. • Developed Angular components for context ingestion (drag-and-drop upload, content preview, confirmation step) that let users validate extracted context before triggering generation, reducing errors in output. Extended the pipeline to support Azure OpenAI.

Education

Bachelor of Technology - BTech, Electronics and Telecommunications Engineering, Walchand Institute of Technology, Solapur