Skip to content
Open to workBengaluru, India

Full stack engineer

I design and build full-stack systems.

01Work

BigHaat

Software Engineer

Dates
Since Nov 2024
Place
Bengaluru
Award
Shining Star, Innovation & Initiatives

Owned solo

The company CRM, start to finish

Requirements, architecture, interface, and build. Several departments run on it. It replaced software the company had been licensing.

90%Operating cost removed
1Engineer on it

Set the precedent

The first tests in the codebase

I found and fixed security issues on the main storefront, then wrote Jest suites to hold them. No repository here had a test in it before.

Moved the org

AI, from argument to practice

I backed it before it was policy, then built the CRM with it from scratch. Adoption went company-wide. My dev loop now ends with a Playwright pass. The agent verifies each finished feature end to end before I review it.

Also shipped

Container tracking, field sampling, a GTM site

Delivery stages and shipment status B2B customers can check instead of asking. The sampling module of an agricultural traceability app, used in the field. A second website for go-to-market.

TypeScript
Next.js
Nest.js
MongoDB
PostgreSQL
Express
React Native
Jest
Docker
Azure

Education

B.Tech, Computer Science & Engineering

Guru Gobind Singh Indraprastha University

CGPA 9.0 / 10.02020 to 2024
02Projects
2026

Argus

AI research copilot

Argus research briefing for Zomato, showing the objective, the completed research run and the generated report

An agent that researches a company on its own and writes a structured sales briefing, with a citation behind every claim it makes.

  • LangGraph workflows where the agent critiques and revises its own output before returning it.
  • Grounded chat over the sources it gathered, with resumable SSE streaming and MongoDB checkpointing so a run survives a dropped connection.
  • FastAPI on AWS App Runner, Next.js on Amplify, images in ECR, shipped by GitHub Actions.
LangGraph
FastAPI
Next.js
MongoDB
AWS
Glass Council home screen, showing the three-stage process and the four active council models

A web app that sends one prompt to four frontier models, has them review each other's answers anonymously, then has a single arbiter model write the final response from the rankings.

  • Queries GPT-5.1, Gemini 3 Pro, Claude Sonnet 4.5 and Grok 4.20 in parallel.
  • Three stages: independent answers, anonymous peer review and ranking, then synthesis by the arbiter.
  • Reviews are anonymous so the ranking can't favour a model by name.
Next.js
TypeScript
Streaming
Multi-LLM

Also

2026

Pericles

Productivity workspace

A productivity workspace with task management, encrypted notes and focus tracking in one interface. Next.js and NestJS in TypeScript, MongoDB behind it, deployed on AWS.

Next.js
NestJS
TypeScript
MongoDB
AWS
2025

Instagram Extension

Chrome extension

A privacy-first Chrome extension with 7,000 weekly users and a 4.7 rating. Processing runs on-device, nothing is transmitted, and rate limiting is built in.

JavaScript
Chrome APIs
Local-first
03Certification
Nov 2024

Machine Learning Specialization

DeepLearning.AI & Stanford Online

3 courses · Andrew Ng

View on Coursera
Coursera Machine Learning Specialization certificate awarded to Heeransh Siddhu, DeepLearning.AI and Stanford Online, November 2024

Supervised learning end to end, across linear regression, logistic regression and neural networks.

Plus the practice around them: cost functions and gradient descent, regularisation, the tradeoff between bias and variance, and reading a model's errors to decide what to fix next.

04Toolkit

Languages

TypeScriptJavaScriptPythonJavaSQLHTMLCSS

Frameworks & Libraries

ReactNext.jsReact NativeNest.jsExpress.jsNode.jsFastAPIRedux ToolkitTailwind CSS

AI & Machine Learning

LangGraphAgentic workflowsLLM APIsTensorFlowscikit-learnNumPy

Databases & Infrastructure

MongoDBPostgreSQLDockerAWSAzureGitHub ActionsGit