PARTNER & COO, GARIBZZ — CURRENTLY BUILDING

Arpan
Khan.

Partner. Engineer. Builder.

Building products that solve real problems through software, systems, and thoughtful engineering — from distributed backends to the businesses they power.

Scroll
01 — About
Arpan Khan

I've always been more curious about how things work than what they claim to do. That curiosity is what pulled me from writing my first scripts into architecting systems, and eventually into starting a company around them.

As Partner & COO of Garibzz, I spend my days moving between two worlds — the engineering room, where I design backends that need to hold up under real load, and the operating room, where I make the calls that keep a startup moving. Neither comes with a manual, which is exactly why I like it.

I care about systems that don't break quietly — caching layers that fail gracefully, queues that never lose a job, authentication that doesn't cut corners. Good engineering, to me, is invisible: the user just sees a product that works.

I'm still learning, deliberately and constantly — new architectures, new failure modes, new ways AI changes what a small team can build. I write about most of it, because the fastest way to actually understand something is to try to explain it clearly.

Role Partner & COO, Garibzz
Focus Full Stack Development
Focus System Design & Backend Engineering
Focus Artificial Intelligence
Approach Scalable Architecture, Product Thinking
Currently
distributed systemscachingLLM toolingstartup opsAPI design
"People may never know my name.
They don't have to.
If the code I leave behind continues to help someone,
that's enough."
— On why I build

Selected work — products and systems built end to end.

A mix of production startups, system-design explorations, and applied AI — each one chosen to solve a real, specific problem.

Garibzz

A modern multi-vendor ecommerce platform built to handle real transactions, real vendors, and real scale — from checkout to fulfillment.

Node.jsReactNext.jsMongoDBRedisDockerBullMQCashfree

Challenge

Support many independent vendors on one platform without a single vendor's traffic or errors taking the whole system down, while keeping checkout fast and payments reliable.

Architecture

Service-oriented Node.js backend, Redis for session and catalog caching, BullMQ-driven job queues for order processing and notifications, all containerized with Docker.

Results

A stable multi-vendor checkout flow with Cashfree payment integration, background order processing that survives spikes, and a catalog that stays fast under load.

Lessons learned

Queues turn unpredictable traffic into predictable work. Caching is a product decision as much as a technical one — it changes what "fast" feels like to a vendor.

System Design

Open Source SaaS Platform

A self-hostable SaaS starter exploring multi-tenant architecture — isolated tenant data, role-based access, and a billing layer designed to be forked and extended, not just demoed.

Multi-tenantRBACPostgreSQLDocker
Applied AI

AI Projects & Tooling

A set of applied AI experiments — LLM-backed internal tools, prompt pipelines built on OpenRouter and Ollama, and computer vision utilities built to solve narrow, real problems well.

LLMsOpenRouterOllamaComputer Vision

How it's built — the systems behind the products.

Engineering is the actual product. These are the pieces I reach for most, and how they fit together under real traffic.

Authentication

Session and token-based auth, built with clear failure modes and no shortcuts on token handling.

Distributed Systems

Services that stay correct and available when a dependency slows down or fails outright.

Caching & Redis

Redis for hot-path reads, rate limiting, and session storage — cache invalidation as a first-class concern.

Docker

Every service containerized for reproducible builds, from a laptop to a production cluster.

Queues & BullMQ

Background jobs for anything that doesn't need to happen in the request-response cycle.

Scalability & Performance

Designing for the traffic pattern a system will actually see, not a hypothetical worst case.

System & API Design

Clear boundaries between services, and APIs designed to be predictable before they're documented.

Cloud & AI Integration

Shipping to Vercel, Render and Cloudflare, with LLMs wired in where they genuinely help the product.

Toolkit — what I build with.

The tools I reach for by default, grouped by where they sit in a product.

Backend

Node.js
Express
FastAPI
Python
MongoDB
PostgreSQL
Redis
Docker
BullMQ

Frontend

React
Next.js
TypeScript
Tailwind CSS

DevOps

Linux
Git
GitHub Actions
Cloudflare
Vercel
Render

AI

OpenRouter
Ollama
LLMs
Prompt Engineering
Computer Vision

Business

Product Strategy
Brand Building
Startup Operations

Timeline — the path so far.

Engineering milestones, learning milestones, and the moments that turned into Garibzz.

Present

Partner & COO, Garibzz

Running engineering and operations for a multi-vendor ecommerce platform — from system architecture to the day-to-day decisions that keep a startup moving.

Backend & Systems

Went deep on system design

Moved from building features to designing systems — queues, caching layers, and services built to survive real, unpredictable traffic.

Applied AI

Started building with LLMs

Began integrating language models into real products and internal tools, treating prompt design and evaluation as an engineering discipline.

2020 — Present

Freelance Developer & Consultant

Working with clients to build custom web applications, robust backend architectures, and AI integrations. Taking ideas from concept to production-ready software.

Foundations

Learned to build, then to ship

Full stack fundamentals first — then the harder lesson: how to take something from a local script to a product someone else depends on.

Get in touch

Let's build
something.

Open to conversations about engineering, products, and interesting problems worth solving.

hello@arpankhan.online