Work
Four years of client platforms, company products, and one of my own.
Sole or lead infrastructure engineer on every one of these, and lead backend developer on most. Client and company names are left out on purpose. The decisions are mine to talk about, the names are not. The one at the bottom of that list is mine, so it gets its real name.
- 01
4 yrs
one employer, plus my own product on the side
- 02
Millions
requests a day on the largest platform
- 03
2
container platforms in production, EKS and ECS
- 04
1
product that is entirely mine, shipped solo
- Client platform90,000+ locations, 500k+ records
Consumer Travel Platform
Global discovery and data platform running on Kubernetes.
travel.architecture
- Ran the whole EKS environment: cluster design, node groups, autoscaling and rollout strategy, with infrastructure defined in AWS CDK.
- 90,000+ locations and over 500,000 records behind it, with Elasticsearch doing geo and full-text search and Redis holding API responses under 100ms.
- GitOps delivery through ArgoCD, so a bad release was one reverted commit rather than a manual scramble.
- CloudFront caching plus query tuning took throughput and page speed to roughly three times what it was.
- Client platform100,000+ users
Fintech Trading Platform
Live stock tournaments, leaderboards and portfolios on a real market data feed.
fintech.architecture
- Ran the microservices behind it: the main API, scheduled jobs, a leaderboard service, and a Socket.io service pushing live prices from a real-time market data feed.
- Split into separate task definitions so the socket service scales alone. When markets open and connections jump, the API and batch jobs stay flat.
- That containment is the whole design. One hot service exhausting shared capacity is how the platform goes down.
- SQS and Lambda took background jobs and notifications off the request path, which kept the APIs light. 99.9% uptime and deploys around 80% faster after moving to containers and Jenkins.
- My own product4 codebases, 1 person
Anonymous India
A social product I own end to end: Flutter app, NestJS API, public site and an admin console.
anonymous.architecture
- Everything is mine on this one. Product decisions, the data model, the API, the app, the infrastructure and the store releases.
- NestJS on the Fastify adapter with Prisma over PostgreSQL 16, Redis 7 for sessions, rate limits and the Socket.io adapter, so realtime works across more than one API container.
- Auth is device-based rather than email or phone. A hashed device id, JWT signed RS256, refresh tokens that can be invalidated the moment somebody is banned.
- Rate limiting, profanity filtering, reports, bans and an audit log all run server side, because moderation that lives in the client is not moderation.
- Multi-stage Docker build on node:20-alpine running as a non-root user, with Postgres, Redis, the API and Nginx composed together for production.
- Flutter on the front with Firebase for push, crashlytics and analytics, RevenueCat for subscriptions and AdMob underneath the free tier.
- Company product
AI Tools Directory
AI tools directory with prompt-based search, owned end to end.
- Product direction, backend and infrastructure, all mine.
- Search runs on the OpenAI API: describe the problem in plain language and get the tools that fit, rather than guessing at a category.
- Node.js and Express on MySQL, with Elasticsearch search across the catalogue and Redis caching the busiest queries.
- An image resizing pipeline behind S3 and CloudFront, so a content-heavy catalogue still loads fast.
- Also the technical SEO and page speed work, which is the half of a directory that decides whether it gets found.
- Client platform
Digital Business Card Platform
Digital business cards linked to physical NFC products.
- Built the entire backend, and ran both the client relationship and the delivery team through to launch.
- Card creation and editing, QR-based profile sharing, and Apple Wallet passes.
- Paired with physical NFC hardware: metal cards, wristbands and stickers, each mapped to a profile.
- Used the OpenAI API for card recognition on scan, so a photographed card becomes a contact without retyping it.
- Client platform
Industry News Mobile Application
Sector news, regulation updates and a knowledge base.
- Built the backend and APIs behind the mobile app: content delivery, search, and the feed that keeps professionals current on regulatory changes.
- Client platform
Enterprise ERP
Back-office platform for a large industrial group, running on Azure.
- Built the whole backend: employees, attendance with shift logic, leave, equipment, inventory, vendors, purchase, expenses and maintenance, in one dashboard.
- Role based access across admins, HR, vendors and technicians, so each group only sees what it should.
- This one runs on Azure rather than AWS. Blob Storage behind Multer for uploads, Azure Pipelines for delivery.
- It replaced a manual paper workflow. Reporting that used to take a day comes back in minutes.
Want the parts I left out?
Happy to go through any of these properly on a call, including the decisions I would make differently now and the ones that cost me a night. Names of clients stay out of it, the engineering does not.
deepboda18@gmail.com