I build backends
that hold up in production.
I'm a backend-heavy full-stack engineer. I own the backend, database and DevOps at MealPe — an institutional food-services SaaS processing 50,000+ meals a month for 20,000+ users across campuses, workplaces and hospitals.
I'm the sole engineer behind MealPe's backend, database and infrastructure — and I took it on while finishing my Computer Engineering degree. I joined when the platform had around 1,000 users; I own every layer that carried it to 20,000+: the APIs, the schema, the servers, the CI/CD pipeline, and the 3 AM incident calls.
Most of what I do is the unglamorous work that decides whether a product survives real traffic — cutting cloud spend without downgrading the experience, shipping zero-downtime deploys, and keeping queries fast as load climbs. I care less about chasing frameworks and more about systems that stay up, stay fast, and stay cheap to run.
MealPe Online Food Ordering POS Pvt. Ltd.
Passdn Technologies Pvt. Ltd. (Vayu App)
Five areas I've been accountable for in production, not just used on a side project. Each one links to a write-up of the actual decision and what it cost to get wrong.
Designing REST APIs that stay correct under concurrency — explicit order state machines, server-authoritative business rules, and idempotency keys so a retried request never charges twice. I treat the client as untrusted input, always.
Composite indexes over more indexes, database views for read paths, RPC functions to push work down to the database, and pagination introduced before it becomes an emergency. Currently tuning a schema serving 120,000+ queries a day.
The transactional outbox pattern, FOR UPDATE SKIP LOCKED job queues, at-least-once delivery with idempotent consumers, and reconciliation sweeps that repair drift between systems. Usually in plain Postgres rather than a message broker.
Owning the full server lifecycle on AWS: EC2, S3 lifecycle policies, NGINX, SSL, PM2 cluster mode and CI/CD. I cut a production AWS bill by roughly 60% by fixing logging and static-asset serving rather than downgrading instances.
Tenant isolation enforced in middleware so a missing filter fails closed, fine-grained permissions modelled as capabilities rather than role checks scattered through controllers, and subsidy budgets that survive concurrent spending.
Four rules I did not start with. Each one is the residue of an incident.
Anything the client can send, an attacker can forge and a crashed screen can lose. Prices, permissions, order state and dietary rules are computed and enforced server-side. The UI is a view, not an authority.
Webhooks retry, networks partition, and sockets disconnect mid-flight. I design for at-least-once delivery and make the effects idempotent — usually with a unique constraint the database enforces — rather than hoping for exactly-once.
A second system to operate is a permanent cost paid to avoid a temporary problem. Postgres does durable queues, locking and consistency well enough that most teams reaching for a broker did not need one yet.
Every meaningful performance win I have shipped came from a slow-query log or a billing dashboard, not intuition. The corollary: instrument early, because you cannot fix what you never measured.
Patient meal ordering with dietary-compliance rules and HIS integration, where a wrong delivery is a clinical incident rather than a refund. See the MedMeals case study.
Multi-vendor ordering, POS integration and peak-load lunch rushes across campuses and workplaces at 50,000 transactions a month. How that platform scaled.
Corporate subsidy budgets, tenant isolation and role-scoped admin dashboards where one missing query filter leaks another company's data. The authorization model I use.
Razorpay and Easebuzz gateway integration, webhook reliability, and dynamic ride pricing plus ONDC network onboarding from my time on a mobility app. Why payment is my source of truth.
Roast your Mac's bad storage habits, one folder at a time.
Patient meal ordering and dietary compliance, synced to the hospital's HIS.
Connects alumni with their institutions — built and shipped in 36 hours.
NLP that reads resumes and scores them against job descriptions.
Ongoing academic research
LDRP Institute of Technology and Research, Gandhinagar
Completed final year while working full-time as a Software Engineer at MealPe.
GitHub's native stacked pull requests hit public preview on July 30, 2026 — dependent PRs that review as separate layers and rebase server-side when the bottom one merges. I've hand-rolled this workflow with 'depends on #142' comments and rebase cascades for years. Here's how the native version works, what it replaces, and where I'd still be careful.
Client, operator, caterer — one record, three audiences, none trusting the others. Modelling a three-party workflow in Postgres with read-time projections.
A kitchen buys in bags, cooks in grams, counts in kilos. How I modelled a multi-unit inventory ledger in PostgreSQL: one base unit, one lock target, one log.
The hype says throw an AI agent at manual data entry. I automated a menu-onboarding pipeline — hundreds of items, images, matching — with boring deterministic code and a human-in-the-loop for the ambiguous 10%. Here's the pipeline, why an LLM would have been the wrong first tool, and the exact two places a vision-language model does earn its cost.
Poojan Ghetiya is a full stack developer and backend engineer based in Ahmedabad, Gujarat, India. He is the sole owner of the backend, PostgreSQL database, AWS infrastructure and deployment pipeline at MealPe, an institutional food-services SaaS that processes over 50,000 meal transactions a month for 20,000+ registered users across campuses, workplaces and hospitals. He holds a B.E. in Computer Engineering from LDRP Institute of Technology and Research (CGPA 8.51, class of 2025) and won Smart India Hackathon 2024 at the national level.
Both, with the weight on the backend. Poojan Ghetiya works as a full stack developer — he ships React.js and Next.js interfaces as well as the APIs behind them — but the depth of the work is backend and infrastructure: Node.js and TypeScript services, PostgreSQL schema design and query tuning, and the AWS servers, NGINX config and deployment pipeline they run on. If you are hiring for one label, "backend engineer who can own the full stack" is the accurate one.
This is Poojan Ghetiya the Node.js and AWS backend engineer from Ahmedabad, Gujarat, India — GitHub POOJANGHETIYA, LinkedIn /in/ghetiyapoojan, and this site at poojan.technokari.com. He is the backend and infrastructure owner at MealPe and a national winner of Smart India Hackathon 2024. Ghetiya is a common surname in Gujarat and several other software engineers share it, including Laravel, Drupal, Java and .NET developers; this Poojan Ghetiya works specifically in Node.js, TypeScript, PostgreSQL and AWS.
His core stack is Node.js and TypeScript on the backend, PostgreSQL and Supabase for data, and AWS (EC2, S3) with NGINX, PM2 and GitHub Actions for infrastructure and deployment. He also works with React.js and Next.js on the frontend, and has integrated payment gateways (Razorpay, Easebuzz), hospital HIS systems and the ONDC network.
He scaled a multi-tenant food-services SaaS from roughly 1,000 to over 20,000 registered users while owning its entire backend, reduced its AWS infrastructure bill by about 60% without downgrading service, moved it to zero-downtime deployments using PM2 cluster mode and NGINX, and built its hospital product MedMeals including dietary-compliance rules and HIS integration.
Yes. He takes on freelance projects, consulting engagements and full-time roles, with a focus on backend architecture, PostgreSQL performance, production reliability and cloud infrastructure. The fastest way to reach him is to book a 30-minute intro call or email swe.poojan@gmail.com.
He publishes long-form engineering articles drawn from running a production SaaS: payment webhook idempotency, the transactional outbox pattern in PostgreSQL, fine-grained multi-tenant RBAC without a policy engine, real-time tracking with Socket.IO, PostgreSQL indexing and query optimisation, AWS cost reduction, and zero-downtime deployment with PM2 and NGINX.
Building a SaaS and need someone to own the backend end-to-end? Or just want to talk shop about Node.js, AWS or production systems? I'm open to freelance projects, consulting and full-time opportunities.
The fastest way to reach me. Grab a slot in my calendar and we'll talk through your project or backend challenge directly.
🔒 I usually reply within 24–48 hours. Fields marked are required.