Nexus ERP
A multi-tenant SaaS ERP with 20+ modules across inventory, purchasing, manufacturing, accounting, payroll, HR and analytics. The hard part is tenant isolation, permissions, ledger integrity and consistent data rules.
S/01 Full-stack development — Dhaka, Bangladesh
Full-stack development that survives production, not just the demo.
Suhanur Rahman builds web applications and internal systems for teams that need more than a polished front end. The work spans architecture, data models, APIs, authentication, automation and deployment — with an emphasis on systems that stay understandable after launch.
0120+
field research projects informed
0220+
ERP modules in Nexus
0304
core engineering layers
0401
owner across the stack
Live system
Software you can actually operate, not just demo.
01 — The problem
Where the complexity usually starts.
01
The prototype works, but nobody trusts the codebase enough to extend it.
02
Your product has too many manual steps between database, API and users.
03
Different contractors own different layers and no one owns the full delivery path.
04
You need a production-minded developer who can also understand the business workflow.
02 — Engineering approach
From a product requirement to a system you can actually operate.
A typical build starts with the data model and the awkward constraints, not the landing page. The goal is to make the important rules explicit: who can see what, which records can change, what has to remain auditable, and where automation should replace manual repetition.
That has led to work on multi-tenant SaaS products, research platforms, civic technology, publishing automation and document-processing tools. Across them, the recurring problem is the same: information has to move reliably through several layers without losing meaning.
The stack is deliberately practical. TypeScript on the application side, React or Next.js at the interface, Node.js and Express for services, PostgreSQL and Prisma for structured data, with Git, Linux, nginx, PM2, Cloudflare and VPS infrastructure where self-hosting makes sense.
03 — Delivery model
The useful unit of work is a solved problem, not a pile of tickets.
01
Map the constraint
Start from the data model and the awkward rules, not the landing page.
02
Build the smallest useful slice
Ship a working vertical slice and keep every decision visible.
03
Test against real behaviour
Validate on real users, edge cases and messy production data.
04
Ship, document and improve
Leave maintainable code and context for the next engineer.
04 — Selected work
Systems with real operational constraints.
Bidyut-Prohor
A civic data platform built around anonymous outage reporting, real-time aggregation and automated verification. It treats an untrusted public feed as a data-quality problem rather than assuming every report is true.
PDF → Markdown
A document pipeline for Bengali PDFs, combining scanned-page OCR with Bijoy-to-Unicode conversion so archival material becomes searchable, structured and easier to reuse.
Publishing automation
Automation around official social APIs, scheduling, content formatting and AI-assisted production — removing the repetitive step without creating a fragile black box.
05 — For international teams
A Bangladesh-based engineering partner for remote product work.
For an overseas founder, agency or research organization, the useful distinction is not the time zone on the map. It is whether a developer can communicate constraints, work asynchronously, document decisions and take ownership after the first deployment.
That is the kind of engagement this portfolio is designed to support: scoped technical work with a visible deliverable, a clear handover and the option to stay involved for iteration or operations.
Projects are a particularly good fit when software, data and research meet: internal tools, workflow automation, research platforms, evidence-processing systems and products where reliability matters more than novelty.
06 — Start here
Discuss the project Looking for a full-stack developer in Bangladesh?
Send a short description of the problem, the current stack and what needs to exist at the end of the engagement. A useful first conversation can usually establish whether the work fits.
07 — Related workAll services