Prashant Yadav.
Portrait of Prashant Yadav
Prashant YadavOpen to lead AI & engineering roles

Lead AI & Full Stack Engineer · Agentic AI · Pune, India

I build agentic AI systems, and the platforms they run on.

11+ years designing enterprise-scale, cloud-native platforms for banking, fintech, e-commerce and manufacturing. Right now I'm leading an Agentic AI platform that serves 500+ engineering leaders through Microsoft Teams, built on LangGraph, FastAPI, pgvector and AWS Bedrock.

Get in touchLinkedInGitHubRésumé (PDF)

Move your cursor: the background is a live nearest-neighbour search.

trace prashant.career11y 4m · running
  1. careerJun 2015 → now
  2. Habilelabs3y 8m
  3. Polls Guru · Hiretale
  4. Wipro2y 8m
  5. HSBC loan platform
  6. Coforge5y · now
  7. Agentic AI platform · MatcoTools

Selected work

Production systems, measured by what changed for the business

Client code is private, so each project is described by the problem, the approach and the result.

Enterprise platform · Coforge · Current

Enterprise Delivery Intelligence Platform

Delivery information for engineering teams lives across Jira, GitHub and Confluence. I'm leading the architecture and delivery of an Agentic AI platform in Microsoft Teams that answers engineering leaders' delivery questions from those sources, with citations they can check.

  • Python
  • FastAPI
  • LangGraph
  • Multi-agent systems
  • RAG
  • PostgreSQL
  • pgvector
  • Redis
  • AWS Bedrock
  • AWS ECS
  • SQS
  • S3
  • Terraform
  • 500+engineering leaders served
  • 50+teams on the platform
  • ~50%less manual delivery investigation

Enterprise platform · Coforge · Current

Enterprise Delivery Intelligence Platform

Delivery information for engineering teams lives across Jira, GitHub and Confluence. I'm leading the architecture and delivery of an Agentic AI platform in Microsoft Teams that answers engineering leaders' delivery questions from those sources, with citations they can check.

Results

  • 500+engineering leaders served
  • 50+teams on the platform
  • ~50%less manual delivery investigation

How it flows

  1. 01Askquestion in Teams
  2. 02PlanLangGraph agents
  3. 03Retrievehybrid RAG, pgvector
  4. 04Verifysources + citations
  5. 05Answercited reply in Teams

Simplified view of how a question is answered. Jira, GitHub and Confluence data arrives through event-driven ingestion (webhooks, SQS, S3, Redis).

What I did

  • Leading architecture and delivery of an enterprise Agentic AI platform serving 500+ engineering leaders across 50+ teams.
  • Designed LangGraph-based multi-agent workflows with source verification and citation mechanisms.
  • Architected hybrid RAG using PostgreSQL + pgvector across Jira, GitHub and Confluence data sources.
  • Built event-driven ingestion pipelines using AWS SQS, S3, Redis and webhooks.
  • Established RBAC, audit logging, observability and security controls for enterprise adoption.
  • Reduced manual delivery investigation effort by approximately 50%.

Tech stack

  • Python
  • FastAPI
  • LangGraph
  • Multi-agent systems
  • RAG
  • PostgreSQL
  • pgvector
  • Redis
  • AWS Bedrock
  • AWS ECS
  • SQS
  • S3
  • Terraform

MatcoTools · Coforge

AI Shopping Assistant & Visual Product Discovery

Technicians often know what's broken but not the part number. I architected a multimodal assistant where they describe what they need or upload a photo of a broken tool or missing vehicle part, and get compatible replacements with technical guidance.

  • Python
  • FastAPI
  • RAG
  • LangChain
  • Pinecone
  • AWS Bedrock
  • HuggingFace
  • React
  • Next.js
  • +15%recommendation conversion
  • Text + photosearch by description or by image

MatcoTools · Coforge

AI Shopping Assistant & Visual Product Discovery

Technicians often know what's broken but not the part number. I architected a multimodal assistant where they describe what they need or upload a photo of a broken tool or missing vehicle part, and get compatible replacements with technical guidance.

Results

  • +15%recommendation conversion
  • Text + photosearch by description or by image

How it flows

  1. 01Askquestion or photo
  2. 02Embedtext + image embeddings
  3. 03SearchPinecone vectors
  4. 04ReasonLangChain + Bedrock
  5. 05Recommendcompatible parts + guidance

Simplified view of the visual search and recommendation flow.

What I did

  • Architected a multimodal AI assistant that lets technicians identify tools and vehicle parts through natural language and image uploads.
  • Developed a visual search workflow that matches photos of broken tools or missing vehicle components against the product catalog using image embeddings and vector search.
  • Built a RAG-powered recommendation engine with LangChain, Pinecone and AWS Bedrock for technical guidance and compatible replacement recommendations.
  • Increased recommendation conversion by 15% through AI-powered search and personalized product discovery.

Tech stack

  • Python
  • FastAPI
  • RAG
  • LangChain
  • Pinecone
  • AWS Bedrock
  • HuggingFace
  • React
  • Next.js

MatcoTools · Coforge

E-commerce and financing platform for the tool trade

A commerce platform serving technicians and distributors across the USA and Canada, with built-in financing so technicians can buy expensive equipment on EMI.

  • Next.js
  • React
  • TypeScript
  • GraphQL
  • MongoDB
  • Microservices
  • AWS
  • Jenkins
  • 2.5 s → 800 mscheckout latency on purchasing APIs
  • +25%financed purchases
  • 150,000+technicians and 2,000+ distributors

MatcoTools · Coforge

E-commerce and financing platform for the tool trade

A commerce platform serving technicians and distributors across the USA and Canada, with built-in financing so technicians can buy expensive equipment on EMI.

Results

  • 2.5 s → 800 mscheckout latency on purchasing APIs
  • +25%financed purchases
  • 150,000+technicians and 2,000+ distributors

What I did

  • Designed and scaled the platform for 150,000+ technicians and 2,000+ distributors across the USA and Canada.
  • Architected microservices and GraphQL APIs for customer onboarding, product catalog management, financing approvals and payment processing.
  • Built financing and EMI management workflows that increased financed purchases by 25%.
  • Optimized critical purchasing APIs, cutting checkout latency from 2.5s to 800ms.

Tech stack

  • Next.js
  • React
  • TypeScript
  • GraphQL
  • MongoDB
  • Microservices
  • AWS
  • Jenkins

HSBC · Wipro · 2019–2021

Loan management for enterprise banking

Enterprise loan management modules for investors, suppliers and buyers inside HSBC's banking systems, built on React, Node.js and microservices.

  • React
  • Node.js
  • Express.js
  • JavaScript
  • REST APIs
  • PostgreSQL
  • Microservices
  • AWS
  • Git
  • −30%loan processing time
  • −25%application response times

HSBC · Wipro · 2019–2021

Loan management for enterprise banking

Enterprise loan management modules for investors, suppliers and buyers inside HSBC's banking systems, built on React, Node.js and microservices.

Results

  • −30%loan processing time
  • −25%application response times

What I did

  • Developed enterprise loan management modules for investors, suppliers and buyers using React, Node.js and microservices architecture.
  • Built and integrated REST APIs supporting loan approval workflows, reducing processing time by 30%.
  • Optimized application performance through frontend and API enhancements, improving response times by 25%.
  • Contributed to AWS-based deployment, monitoring and production support.

Tech stack

  • React
  • Node.js
  • Express.js
  • JavaScript
  • REST APIs
  • PostgreSQL
  • Microservices
  • AWS
  • Git

Earlier projects

Habilelabs · 2015–2019

Polls Guru & Hiretale

Survey platform and job portal

Led development of cloud-based polling, analytics and recruitment platforms for enterprise users, including real-time dashboards and reporting, AWS deployments and CI/CD pipelines.

  • −60%processing effort
  • +25%user engagement
  • −40%page load time
  • 99.9%availability
  • React
  • JavaScript
  • Node.js
  • Express.js
  • MongoDB
  • REST APIs
  • AWS
  • CI/CD

Experience

Three companies, one thread: products that hold up under real traffic

  1. Oct 2021 – Present

    Tech Lead · Coforge

    Leading an enterprise Agentic AI platform for 500+ engineering leaders, plus the AI shopping assistant and the e-commerce and financing platform for MatcoTools.

  2. Feb 2019 – Oct 2021

    Senior Project Engineer · Wipro

    Loan management modules, REST APIs and approval workflows for HSBC.

  3. Jun 2015 – Feb 2019

    Software Engineer · Habilelabs

    Led development of Polls Guru and Hiretale on React, Node.js, MongoDB and AWS.

Skills

What I work with

Hover or tap a skill to see where I've used it. used in a project on this page

AI & GenAI
Backend
Frontend
Cloud & DevOps
Databases
Architecture & design

Certifications & education

Credentials

  • AWS Certified Developer – AssociateAmazon Web Services
  • Associate Cloud EngineerGoogle Cloud
  • Generative AI with Large Language ModelsDeepLearning.AI and AWS
  • Appathon winnerHack Day Codefest, 24-hour app build
  • B.Tech, Computer ScienceJaipur National University · 2011–2015

Contact

Building something with LLMs?

I'm open to lead AI and full-stack engineering roles. Email is the fastest way to reach me.

prashant.yadav567@gmail.comLinkedInGitHub