Skip to content
Portrait of Abhinav Patidar

Abhinav Patidar

Building Estimo. Open to hard problems.

I build softwarethat runs businesses.

Senior Software Engineer · Systems Builder · Product Engineer

I turn complicated business processes into scalable software systems, products, automation, and intelligent tools.

Role
Senior Software Engineer & AI Engineer
Company
Indus Analytics
Since January 2023
Based
Indore, India
--:-- IST
Flagship
Estimo ERP
Printing & packaging

From problem to product.

What arrives is rarely a spec. It is a plant that quotes in Excel and plans on the phone. The work is turning that into a system that holds.

Real-world problem

  • Excel everywhere
  • Manual calculations
  • Repetitive processes
  • Disconnected systems
  • Data inconsistencies
  • Communication gaps
  • Manual reporting

Engineering

  1. Understand
  2. Model
  3. Design
  4. Build
  5. Automate
  6. Improve

System

  • ERP
  • APIs
  • Database
  • Integrations
  • Web & mobile
  • Automation
  • Analytics
  • AI where useful

Flagship

Estimo

The ERP behind printing & packaging businesses.

Indus Analytics' ERP for printing and packaging manufacturers. Everything from the first enquiry to dispatch and MIS reporting, in one multi-company system. I build and support it end to end: the Next.js costing platform, the C# Web APIs behind it, the SQL Server data model, and the AI layer on top.

This isn't a dashboard project. This is a business operating system. Follow one order through it:

ORDER 2417 · 10,000 folding cartons · example

Stage 01 / 08

Enquiry & Estimation

Material, machine and process costing with per-quantity columns and freight terms.

Cost head5,00010,00025,000
Board 350 GSM2.902.752.60
Printing 4C + varnish1.601.200.95
Lamination0.700.550.45
Die-cut & pasting1.100.650.40
Freight0.500.300.20
Unit cost₹0.00₹0.00₹0.00

Under the hood

Backend
C# Web APIs on .NET Framework 4.8, moving to .NET 10
Frontend
Next.js 16 / React 19, one responsive codebase from 320px up
Data
SQL Server, multi-company and multi-production-unit model
Real-time
SignalR push, audit log and notifications on every action
Mobile
PWA with service-worker push, Capacitor for Android and iOS
Integrations
SAP, Zoho and other ERPs, WhatsApp messaging, Microsoft Graph, AWS S3, Azure
Intelligence
Semantic Kernel assistants calling the ERP's own services

One codebase, every device

Before

DevExtreme, jQuery and VB.NET desktop screens

After

One responsive Next.js application. Module visibility on mobile is a database flag, the DataGrid switches to a card view under 768px, and Capacitor wraps the same build for Android and iOS.

Integrations

An ERP never works alone.

Real plants already run SAP or Zoho, talk to customers on WhatsApp and send everything by email. I connect Estimo to those systems so data moves on its own instead of being typed twice.

EstimoERP COREWhatsAppMessagingSAPERPZohoERP & accountingOther ERPsERPMicrosoft 365Email via GraphAWS S3 · AzureCloud storage

Messaging

WhatsApp

Users message customers on WhatsApp straight from the application, from the screen they are already working on. No copying numbers, no switching to a phone.

  • → Quotations
  • → Order updates
  • → Documents

Estimo · Quotation

QT-0931 · ₹54,500

Send on WhatsApp
CW

Carton Works

Business account · example

QT-0931.pdf

Quotation · 1 page

Your quotation for 10,000 cartons.

11:42

Order SO-1182 is dispatched. Challan DC-0764 attached.

11:43

More systems.

Different problems, same approach. Only projects that exist, with no invented numbers.

Offline-first farm PWA

KhetFlow

DeviceDexieCloudSupabaselast write winssoft deletes
  • Next.js
  • TypeScript
  • Tailwind CSS
  • Dexie (IndexedDB)
  • Supabase
  • PWA
  • Vercel

Tracks every crop cycle of a shade-net farm in Jhalawar, Rajasthan, from seed purchase to final sale.

Sales and client platform

Indus CRM

  • ASP.NET
  • Web APIs
  • SQL Server

Lead, client and dashboard management for multiple live clients, in the same product family as the ERP.

And then I add intelligence.

When traditional software isn't enough, I experiment with AI, agents, LLMs and automation to make systems more capable. AI is integrated into real software systems, not treated as a chatbot demo.

Synthia / ParkBuddy

A router assistant that hands off to specialist Costing, Printing and MIS assistants. Each one calls real backend tools instead of guessing, so a planner can ask a question in plain language and get an answer computed by the same services the ERP uses.

Assistants
General router, Costing, Printing, MIS
How it works
Tool calling into the ERP's own services
Prompts
Strict, for deterministic, no-filler answers
My role
Architecture, prompts, backend tools
  • LLMs
  • Semantic Kernel
  • AI agents
  • MCP
  • Tool calling
  • Automation
  1. User
  2. AI router
  3. Specialised assistants
  4. Business tools
  5. ERP · APIs · Database

The stack, by job.

Grouped by what each part does in a system. Not a count, not a cloud.

Build

  • C#
  • .NET Framework 4.8 → .NET 10
  • ASP.NET Web API
  • Dapper / ADO.NET
  • Background workers
  • xUnit

Interface

  • Next.js 16 / React 19
  • TypeScript
  • Tailwind CSS / Radix UI
  • TanStack Query & Table
  • ECharts dashboards
  • PDF & Excel export

Data

  • SQL Server design
  • Stored procedures
  • Multi-company data models
  • Excel-to-SQL import pipelines
  • SignalR real-time

Automation

  • Python utilities (PyQt6, pyodbc)
  • Playwright / Puppeteer
  • Scripting & internal tools

Integrations

  • SAP
  • Zoho
  • Other ERP connectors
  • WhatsApp messaging
  • Microsoft Graph email
  • REST APIs & webhooks

Intelligence

  • OpenAI tool calling
  • Microsoft Semantic Kernel
  • Multi-assistant orchestration
  • MCP servers & custom agents
  • Claude Code / Codex / Cline

Delivery

  • PWA / service workers
  • Capacitor for Android / iOS
  • React Native / Expo
  • AWS S3 / Azure storage
  • Vercel

Things I build because I’m curious.

Experiments and internal tools. What I tried, what happened, what I learned.

How I think

  1. Understand the business.
  2. Model the system.
  3. Automate what shouldn't be manual.
  4. Add AI where it actually helps.

Experience

Indus Analytics

Senior Software Engineer & AI Engineer

2023 — Present · Indore, India

Education

Bachelor of Computer Applications (BCA)

  • Build and maintain a multi-module ERP / CRM platform for printing and packaging, across manufacturing, planning, inventory, CRM, dashboards and reporting, for multiple live multi-company deployments.
  • Rebuilt the costing, quotation and approval workflows of the ERP as a Next.js 16 / React 19 application over C# Web APIs, serving desktop and mobile from one codebase.
  • Designed and shipped LLM-powered assistants, including the Synthia / ParkBuddy multi-assistant architecture on Microsoft Semantic Kernel with tool calling into backend services.
  • Built an end-to-end carton cost-estimation engine plus the inventory, sales-order and purchase-order modules around it.
  • Developed bulk costing and item-master save workers and an Excel-to-SQL master import tool.
  • Integrated Estimo with SAP, Zoho and other ERPs, so orders, masters and invoices move between systems instead of being keyed in twice.
  • Built WhatsApp messaging into the application, so users message customers directly from the screen they are working on.
  • Implemented SignalR push, AWS S3 / Azure storage, Microsoft Graph email and QR-based inventory modules.
  • Own production support for live applications: root-cause analysis, change requests, deployments and secure code review.

Where this comes from

I grew up in Jhalawar, Rajasthan, finished a BCA and joined Indus Analytics in Indore in January 2023. The first thing I touched was a live, multi-tenant ERP for printing and packaging manufacturers, built on ASP.NET Web Forms and VB.NET. I learned the domain the hard way: by supporting real plants, real quotes and real dispatches.

That domain knowledge is what I build on now. I rebuilt the costing side of that ERP as a modern Next.js and React application on top of C# Web APIs, one codebase serving desktop and mobile. I added AI assistants that call the same backend tools a planner uses, and I wrote the Python utilities the support team runs every day.

I care about one thing above all: the software should match how the business actually works, and stay simple for the person using it.

Got a difficult problem?

Let's build something useful.

This website is also a system.

Built with Next.js 16, React 19 and engineering obsession. Deployed from GitHub main.

© 2026 Abhinav Patidar

Local time
--:-- IST
Status
Online
For technical visitors