Hey, I'm

MADAN HEGDE

I design and build scalable web applications and AI-powered tools, from idea to production.

AI Systems • RAG • Tool Calling • React • Next.js • TypeScript

Currently building scalable workforce systems at PagarBook and AI knowledge infrastructure at Hearly. Also shipping a commissioned AI agent for a sitar maestro.

View My Work ↓Get in Touch

Engineering Highlights

Key technical contributions across architecture, performance optimization, and product development.

From Frontend Lead to Fullstack + AI at Pagarbook

Scope expanded from leading frontend architecture for Pagarbook's 100K+ business workforce platform to fullstack ownership — now designing and shipping the company's first AI integration across Node.js/NestJS in addition to the web stack.

CRA → Next.js: 20% Better Core Web Vitals

Migrated a large production app from Create React App to Next.js, cutting LCP by ~20% through SSR, route-based code splitting, and image optimization.

Built Platform-Wide Component Library (Tailwind + Radix)

Designed and shipped a shared component library using Tailwind CSS and Radix UI, adopted across all product squads to standardize UI development.

30% Faster Builds: Webpack → Vite + Docker

Replaced the Webpack dev pipeline with Vite and containerised builds with Docker, reducing local build times by ~30% and CI cycle time significantly.

Payroll System: 80% Adoption in Month 1

Built the payroll management module from scratch. Reached 80% user adoption within the first month of launch across the existing customer base.

Built RAG-Powered Knowledge Platform with Paying Users

Designed and shipped a full RAG pipeline (pgvector, chunking, embedding, retrieval) powering an embeddable widget and npm package, with paying customers in production.

Designing Pagarbook's First AI Integration

Owning the design and build of the first AI feature at Pagarbook — a tool-calling pipeline over the production backend that lets 100K+ businesses take real actions on workforce data via natural language.

MADAN HEGDE

Software Engineer & Product Builder

I design and build scalable web applications and AI-powered tools, from idea to production.

AI Systems • RAG • Tool Calling • React • Next.js • TypeScript

Tech Stack

Core

ReactNext.jsTypeScript

AI / LLM Systems

RAG PipelinesTool CallingAgent DesignVector Databases (pgvector)EmbeddingsLLM APIs (OpenAI, Anthropic)Streaming

Frontend Architecture

State Management (Redux, Zustand)Component LibrariesPerformance Optimization

Backend

Node.jsNestJSPostgreSQLFirebase

Infrastructure & Tools

DockerGitCI/CD Pipelines

Mobile

React Native

Experience

Nearly 7 years building production web systems, developer platforms, and AI-powered tools.

2025 — Present

Founder & Engineer @ Hearly

Founder

Building RAG-powered knowledge infrastructure for AI systems, with paying customers and an npm-distributed widget used in production.

Knowledge Platform
- Built a RAG pipeline (ingest → chunk → embed → store → retrieve → rerank) backed by pgvector on Postgres. Chose pgvector over Pinecone to eliminate a managed-service dependency and keep latency under 200 ms.
- Shipped an embeddable website widget and published @hearly-knowledgebase/widget on npm; developers integrate in under 10 minutes.
- Connected the knowledge base to LLM chat completions via structured retrieval, powering paying customers in production.
Meeting Copilot (Earlier Product)
- Built audio processing and structured extraction pipelines converting meeting recordings into summaries, tasks, and searchable transcripts.
- Shipped real-time meeting intelligence features using WebSocket-driven UI.

AITypeScriptReactNode.jsnpm

Click to collapse

2022 — Present

SDE 3 @ Pagarbook

Leading the first AI integration at Pagarbook (in development) — designing a tool-calling pipeline over the production backend that lets users take real actions on workforce data via natural language. Expanded scope from frontend lead to fullstack ownership, now shipping across Node.js/NestJS backend in addition to the web stack.

ReactNext.jsTailwind CSSRadix UINode.jsNestJSAITool CallingDocker

Click to expand

2022 - 2022

Frontend Engineer @ Scenes

Worked on interactive social experiences and real-time collaboration features for a creative platform.

ReactTypeScriptWebSocketCanvas API

Click to expand

2020 — 2022

Co-Founder & CTO @ Digiyogi Technologies

Founder

Co-founded a technology startup and built the product from concept to launch.

ReactNode.jsFirebaseReact Native

Click to expand

2019 — 2020

Frontend Engineer @ Learnyst

Built and maintained frontend systems for an ed-tech platform helping creators launch and manage their own teaching businesses.

ReactJavaScriptReduxSCSS

Click to expand

What I'm Building

AI-first products I designed, built, and shipped: from RAG infrastructure to meeting intelligence.

Hearly ecosystem

Hearly

Hearly Knowledge

AI Memory Layer

Hearly Knowledge — AI Memory Layer

A developer-focused RAG knowledge infrastructure that acts as a scalable memory layer for AI systems and chatbots. Connect documents, audio, and video into a structured knowledge base that AI applications can query in real time.

RAG pipeline: ingest → chunk → embed → store → retrieve → rerank
pgvector-backed semantic search (sub-200ms retrieval)
Memory layer for LLM-powered applications
Human-in-the-loop knowledge editing

Developer Integrations

Embeddable website knowledge widget
NPM package for direct integration
API-driven retrieval for AI agents

@hearly-knowledgebase/widget

Next.jspgvectorRAGEmbeddable WidgetsTypeScript

Visit App How I Built It

Hearly

AI Meeting Copilot

Hearly — AI Meeting Copilot

AI-powered meeting intelligence platform that converts meeting audio into structured summaries, tasks, and searchable insights.

Automated meeting summaries
Task extraction from conversations
Interactive Q&A over meeting transcripts

Next.jsAI APIsAudio ProcessingReact Native

Visit App

Independent project

Sitar Maestro

Sitar Maestro AI Agent

Personal AI Agent (in progress)

Sitar Maestro AI Agent — Personal AI Agent (in progress)

In Progress

A personal AI agent built for a sitar maestro — voice and text interaction, contextual memory of compositions, students, and engagements, with tool use for scheduling and content tasks. Solo build, end-to-end.