— Portfolio · Vol. IV · MMXXVIPune, IN · UTC+5:30

I build quietly fastweb products that feellike they were made by hand.

Sagnik Ghosh — fullstack engineer working at the seam between thoughtful interfaces and the systems that power them.

See selected work Start a project
Currently
SDE @ ZenTrades.ai
AWS Lambda, React Native, AI recommendation systems
Stack
React · Node.js · AWS
Fullstack across web, mobile, and cloud infra
Education
VIT Vellore · Gold Medal
B.Tech CSE, highest scoring student in stream
01 — About

Driven by curiosity,
powered by craft.

A short note about how I work and what I care about.

Fullstack engineer who graduated top of class at VIT Vellore and has been shipping production code at ZenTrades.ai ever since.

I work across the whole stack — React and React Native on the front, Node.js and Python on the back, AWS and Docker holding it together. Recent work spans architecting a parallel AWS Lambda pipeline that unblocked a market expansion, optimizing Puppeteer-based PDF generation to cut memory usage by up to 80%, and integrating AI recommendation systems that hit 87% accuracy in production.

I care about measurable outcomes, clear communication, and code that the next person can change without a call to me. Available across timezones; at my best on small teams moving fast.

collaborationopen communicationships fastmetric-driven
Stack — 2026v.4.2
Frontend
Next.jsReactReact NativeTypeScriptTailwind
Backend
Node.jsExpress.jsDjangoPython
Database
MySQLMongoDBRedis
Cloud
AWSDockerNginxElasticSearch
AI / Tools
LangChainGitLinux
03 — Experience

Where I've shipped.

Two roles, one company. Different chapters of the same story.

Software Development Engineer

current
ZenTrades.aiPune
  • Architected parallel execution via AWS Lambda and Google Apps Script, cutting document generation to under 5 minutes and unblocking the Canadian market launch.
  • Refactored Puppeteer-based PDF generation to a single browser instance, reducing memory usage by 50–80% and CPU load by 30–60%.
  • Shipped a React Native app from scratch, streamlining inspection-to-invoice workflows and cutting post-job processing time to under 7 days.
  • Integrated AI-driven recommendation systems into estimate generation, achieving 87% accuracy in automated citation validation.
  • Refactored the scheduler module to eliminate redundant logic, cutting page load times by over 50%.

Associate Developer

ZenTrades.aiPune
  • Built saved filter preferences with React and Material UI, reducing user interaction steps by over 75%.
  • Designed a scalable data schema and RESTful APIs for efficient storage, retrieval, and seamless frontend-backend integration.
04 — Approach

How a project actually goes.

No mystery process. Three phases, repeated until it ships.

Phase 01

Map the territory

I start by understanding the actual problem — users, constraints, success metrics. Site structure, navigation, and content come out of that, not before it.

Phase 02

Build in public

Cue the lo-fi. I work in tight loops — small, shippable slices — and keep you in the thread the whole way. No surprises at delivery.

Phase 03

Ship & sharpen

From approved design to production code, with measurable checks on performance, accessibility, and the small interactions that make a product feel alive.

Let's build something

Got a projectthat deservesreal care?

Send an email
sagnikghosh18@gmail.com