About
I'm a backend software engineer with 3+ years building and deploying RESTful APIs and web applications in Python and Django. My focus is systems that have to behave correctly when things go wrong at the same time — two requests racing for the last unit of stock, a webhook delivered twice, a payment provider timing out mid-transaction. I care more about a system being defensible under scrutiny than about it looking finished.
Outside of the day job, I write about the reasoning behind the backend decisions I make — why a row lock over optimistic concurrency, why a Redis pre-check in front of Postgres, why a circuit breaker earns its complexity at a given scale — because I think the "why" is the part worth being able to defend.
Experience
Backend Developer (Python / Django)
PiHex Labs
Dec 2024 — Present
- Designed and built a high-concurrency checkout and order-management system using DRF — cart validation, pessimistic stock reservation to prevent overselling, and Stripe webhook-driven fulfillment.
- Built a low-latency real-time order-status engine on Django Channels, ASGI, and WebSockets, with Redis as the channel layer, pushing live updates from Celery tasks to connected clients.
- Resolved race conditions in cart and inventory updates using PostgreSQL row-level locking (
select_for_update) and DB constraints, keeping data integrity under concurrent writes. - Decoupled the request/response cycle from heavy work (fulfillment, email) via Celery + Redis, with
acks_lateand idempotent tasks for crash-safe processing. - Cut PostgreSQL query execution time by 30% through targeted B-Tree indexing and ORM-level
select_related/prefetch_relatedoptimization. - Bridged Firebase (Google/OAuth SSO) to local JWT issuance (SimpleJWT) with refresh-token rotation and blacklisting.
- Reached 90% test coverage with PyTest —
unittest.mockto isolate Stripe webhook logic,CELERY_TASK_ALWAYS_EAGERfor deterministic async tests.
Backend Developer (Python / Django)
Singh International Pvt Ltd
Aug 2023 — Nov 2024
- Architected a cloud-native AWS production environment inside a custom VPC — public/private subnets, Security Groups, and Internet Gateways controlling traffic flow.
- Deployed a stateless app on EC2, offloading state to managed RDS (PostgreSQL) and ElastiCache (Redis) in private subnets, with static/media assets on S3.
- Containerized the stack with Docker and Docker Compose for parity across local, staging, and production.
- Built CI/CD pipelines in GitHub Actions for test and deploy, injecting secrets under the principle of least privilege.
Featured work
Reservly
A high-concurrency reservation / flash-sale system, not another to-do e-commerce clone.
Built to answer one question under real scrutiny: what actually happens when hundreds of buyers hit "buy" on the same last unit of stock at the same instant? The storefront is a thin demo surface — the actual point is proving correct behavior under concurrent load, with real load-test numbers and a recorded proof, not a claim.
- Overselling prevention — Redis pre-check absorbs a thundering herd cheaply; a Postgres row lock (
select_for_update) is the actual arbiter. Proven with 20 concurrent threads in a test and a 150-user Locust run against the real HTTP stack. - Reservation TTL — abandoned checkouts auto-release stock via a Celery beat sweep, race-tested against a concurrent webhook confirming the same order.
- Idempotent Stripe webhooks — a DB unique constraint on the event ID, not a check-then-act flag, so a redelivered event can never double-process.
- Real-time order status — WebSocket push via Django Channels, firing only on genuine state transitions, never on an unrelated field save.
- Circuit breaker on Stripe — fails fast and degrades gracefully instead of hanging the whole checkout path when a downstream dependency is slow.
- RAG-powered shopping assistant — Gemini function-calling over a pgvector-backed product index, with graceful degradation when the model's quota is exhausted.
Full write-up — including the load-test numbers, the two-tabs-racing demo GIF, and the reasoning behind every hard decision — is in the project README.
Skills
Languages & Core
Python, JavaScript, SQL, C++, OOP
Web
Django, Django REST Framework, React, RESTful API design, HTML5, CSS3
Cloud & DevOps
AWS (EC2, S3, RDS, VPC, ElastiCache), Docker, Docker Compose, Nginx, Gunicorn, GitHub Actions, CI/CD
Data & Messaging
PostgreSQL, Redis, Celery, Django Channels, WebSockets
Auth & Payments
Firebase, JWT, OAuth 2.0, Stripe API
Testing & Tools
PyTest, Git, unittest.mock
Education
Netaji Subhas University of Technology (NSUT Delhi)
B.E., Electronics and Communication
August 2022
Get in touch
Open to backend engineering roles — Django, Python, distributed systems, and anything that involves making a system provably correct under load.