Four years building the systems behind multi-tenant SaaS, real-estate portals and medical education — REST APIs, asynchronous data pipelines, and the deployments that keep them running. Currently senior backend developer at AAJneeti Connect, Noida.
4 years
building production backends
30+
projects shipped
200+
clients on one analytics platform
100+
ad accounts synced on schedule
Six projects in more detail. The full list of work follows below.
Multi-tenant marketing analytics for 200+ clients across CPL, hybrid and retainer billing. Client spend monitoring, allowed-budget caps with grouping, campaign manager performance, and daily ad-account funding reconciliation.
Scheduled ingestion across 100+ Meta advertising accounts, each with its own credentials. Error isolation at the credential boundary means one failing account never aborts the run.
Internal multi-tenant CRM built on an open-source backend. I worked on it alongside the team — backend features, feature direction, and guidance to the frontend on architecture and implementation.
One Django backend behind four enrolment flows — fellowships, diplomas and masterclasses, country-priced memberships on Razorpay, conference registration priced by tier and date window, and webinars — plus a content layer and a generic-foreign-key SEO layer across all of it.
Complete Django REST Framework backend for a real-estate platform, consumed by a React frontend.
Web presence for a New Delhi litigation firm. Practice areas, partner profiles, the client wall and a monthly newsletter archive with downloadable PDFs all sit in one codebase, and the whole site opens behind the Bar Council of India disclaimer that Indian law firms are required to show.
Thirty-plus built; twenty-six of them listed here, across SaaS, real estate, healthcare and education. See them as cards
The same six steps on every project, whether it is a one-page portal or a platform billing a hundred clients three different ways.
Work starts from written requirement documents rather than a verbal brief, so the modules are specified before any code is written. If a rule is not written down, it gets asked about, not guessed.
The schema is where the real decisions live. Four enrolment flows that differ in eligibility, payment and record-keeping still belong in one coherent model — without a dozen nullable columns or four copies of the same logic.
DRF endpoints standardised on a single {success, message, data, meta} envelope, errors included, so a frontend never has to write a transformation layer and adding a module never adds a special case.
Anything slow or recurring moves to Celery and Celery Beat, with the error boundary drawn around the smallest unit of work so one failure costs that unit and not the whole run.
N+1 queries removed by computing metrics once per client before serialisation, relations pulled with select_related and prefetch_related, bulk writes, and Redis in front of the heaviest reads with a TTL and an on-demand bust. Tests before deploys, not after: anything touching untrusted input gets a malicious-input test before it ships, and the live system gets checked with real requests afterwards.
Gunicorn behind Nginx on DigitalOcean, with the Celery worker and Beat managed as Systemd services so they come back with the machine and are operated the same way as everything else on the box.
Jun 2024 – Present
Jun 2023 – May 2024
May 2022 – May 2023
BCA, Chaudhary Charan Singh University, Meerut — 2019–2022
Python, SQL, JavaScript
Django, Django REST Framework, REST API design, Celery, Celery Beat
PostgreSQL, MySQL, SQLite, Redis
DigitalOcean, Gunicorn, Nginx, Systemd, Linux, Git
Meta Marketing API, webhooks, CRM integrations
HTML5, CSS3, Bootstrap, JavaScript
Education platforms: course catalogs, enrolment flows, memberships, and conference registration.
I build SaaS backends in Python and Django. Four years of it — most recently a multi-tenant marketing-analytics platform that pulls from 100+ Meta ad accounts, serves 200+ clients and bills them across three different models. On systems like that the hard part is rarely the feature list. It's the seams: tenants sharing one schema, a third-party API that fails halfway through a run, and money that still has to reconcile at the end of the month.
So my work sits in backend architecture. One predictable response shape across the whole API, so nothing downstream needs a translation layer. Data models that still hold up three features later. Asynchronous pipelines that isolate failure per credential, so one bad account never takes down a run. The PostgreSQL and Redis work that keeps reporting endpoints fast as the row count grows. And the deployment itself — Gunicorn, Nginx and Systemd-managed Celery workers, which I run rather than hand off.
In the office that mostly means backend: the frontend team builds and updates the interface, and I step in on the parts they can't take further. On freelance work I handle the whole thing myself — frontend, backend and deployment. The work spans SaaS, real estate, healthcare, education and e-commerce.
Noida, India. Open to conversations about backend work.
