- Home
- Tech stack
TECHNOLOGY & EXPERTISE
Our technology stack — and why we chose each piece.
We're specialists, not generalists. Twelve years in the Python ecosystem means we've already made the expensive mistakes on somebody else's project.
Backend & core
Django
Everything we build runs on Django. It ships with an admin panel, an ORM, migrations and a security model that has been attacked by the internet for twenty years. For business software that is a two-month head start over assembling a stack from parts.
Python
One language across web, data, automation and desktop. It means the same engineer can follow a problem from a browser form to a background worker.
Django REST Framework
Serialisers, permissions and versioned endpoints out of the box — so the Android app and the web client can share one backend safely.
Celery
Anything slow or retryable leaves the request cycle: imports, reports, payment callbacks, notification fan-out.
Flask
For small single-purpose services where a full framework would be overhead.
Databases & search
PostgreSQL
Real transactions and real constraints. If your system touches stock or money, this is not a preference, it is a requirement.
Redis
Cache, queue broker, rate limiter and session store. Usually the cheapest performance win available.
Cassandra
For write-heavy, time-series-shaped data at volumes where PostgreSQL would need sharding.
Apache Solr
Faceted search over large catalogues when PostgreSQL full-text stops being enough.
Dgraph
Graph queries where relationships, not rows, are the thing you are searching.
MySQL / MongoDB
Because inherited systems exist and someone has to maintain them competently.
Cloud & DevOps
AWS
ECS Fargate, Lambda, RDS, S3, CloudFront, ElastiCache, ALB. We are certified in it and we have run production microservices on it.
Terraform
Infrastructure becomes reviewable in a pull request instead of living in one engineer’s memory.
Docker
The same image runs in CI, staging and production. No environment drift, no “works on my machine”.
GitHub Actions
Build once, promote through environments, roll back in one click.
Nginx
Reverse proxy, TLS termination and micro-caching — often the difference between a slow site and a fast one.
Ansible
For the servers that are not containers yet, configured repeatably rather than by hand.
DigitalOcean
For small predictable Django workloads it is cheaper and simpler than AWS, and we will say so when it is.
CloudWatch / Sentry
Alarms on what matters, stack traces when it does not.
Mobile
Android (Kotlin/Java)
Native, because an app that must work on a four-year-old phone on 3G is the wrong place for a heavy cross-platform runtime.
Play Store release engineering
Staged rollouts and crash gates, so a bad build reaches one percent of users instead of all of them.
Firebase Crashlytics
Crash triage by device and Android version, so fixes get prioritised by real impact.
Frontend
Bootstrap 5
A grid and component set your next developer will already know. Boring on purpose.
HTMX
Interactive pages without shipping a single-page app you would then have to maintain forever.
JavaScript / AngularJS
Vanilla for new work; AngularJS only where we are maintaining something that already exists.
AI & data
LLM agent frameworks
Tool-calling agents with strict output schemas, a human review step, and cost per run instrumented from day one.
PyTorch / TensorFlow
Used in a bank’s eKYC pipeline for document and face verification, not for demos.
Pandas
Data cleaning and migration work — the unglamorous half of every ERP project.
Scrapy
When the input is the open web, at volume, reliably, and politely.
Payments & integrations
bKash · Nagad · SSLCommerz
All three integrated in production, including the reconciliation and timeout cases the documentation does not cover.
Telco billing gateways
A recurring billing integration handling roughly 300,000 payments a day.
Core banking APIs
Integrated for an ML-based eKYC account-opening flow at a commercial bank.
Stripe
For international clients billing outside Bangladesh.
We don't just use AWS. We're certified in it.
Certifications are the floor, not the ceiling — but they let a technical evaluator stop guessing.
- AWS DevOps: Operate and Monitor
- AWS DevOps: Release and Deploy
- GOOGLE IT Automation with Python
- PYTORCH Deep Learning with PyTorch
- UX UX Design Foundations
Architecture showcase
Sanitised diagrams from real systems. For a technical evaluator this is the highest-credibility asset on the site — worth drawing properly rather than sketching.
ECS Fargate microservices
The estate behind the 60%-faster-deploy engagement, in Terraform.
Django + Redis + Celery pipeline
The shape of the 300,000-payments-a-day billing integration.
Offline-first mobile sync
How the 400K-install app stays usable on 3G and 1GB of RAM.
Not sure which stack fits your problem?
That's what the free call is for.
