Django the way it should be in 2026.
Async views where they pay, HTMX where it fits, and the ORM tuned for the data shape you actually have. Not the textbook version.
Django is still the right answer. The patterns have moved.
Django remains one of the fastest stacks to ship a complete product on. What has changed in 2026 is the patterns: async views are stable, HTMX has replaced the React-everywhere reflex for internal tooling, the ORM has caught up enough that raw SQL is rarely the answer. Most Django shops we audit are running 4-version-old patterns on a stack that has moved on. We help update the parts that matter without rewriting the parts that work.
What changes when a Metafic pod is in your repo.
Async views where they pay back
Outbound API calls, fan-out work, anything I/O-bound. We do not async-ify CRUD just because we can.
HTMX for internal tooling and partial updates
Less JS to write, less JS to ship, faster to build. Django Templates render the partials, HTMX swaps them. We pick HTMX for the admin and the dashboard, not for the public site.
ORM tuning as a discipline, not a heroics moment
select_related, prefetch_related, only(), defer(), and the @cache_property tax. We add the django-debug-toolbar in dev and the N+1 catch in CI.
Celery + Redis tuned to actual load
Connection pool sizing, queue-specific concurrency, dead-letter handling. Or we migrate you to django-q2 or Procrastinate if Celery is more weight than value.
Storage and media handled at the right layer
django-storages with S3/R2, signed URLs, lifecycle policies. We do not let user uploads live next to application code.
Who is on the pod for this work.
Pods scale up from here for Enterprise engagements.
Has shipped Django in production for 8+ years. Has opinions on signals (mostly: avoid them).
5+ years Django, comfortable with HTMX, async, and the deeper ORM internals.
pytest-django, factory-boy, and integration tests against the real database.
Tuned to generate model migrations, propose ORM optimisations from query traces, and write factory definitions.
The bugs that bite this stack.
Custom managers that hide N+1
A clean .objects.published() helper that fans out to 200 queries when used in a list. We trace and fix.
Signal handlers doing too much
A post_save signal that fires emails, updates caches, and writes audit logs. We move them to explicit service calls.
Migrations that lock tables in production
Adding a non-null column without a default on a 50M-row table. We use staged migrations with django-pgschema or pg-osc.
Settings.py that has grown to 800 lines
The classic. We extract environment-specific config and remove the dead toggles.
Honest about scope.
We will not migrate you from Django to FastAPI just because FastAPI is faster on micro-benchmarks. The constraint is rarely framework throughput in a Django shop. The constraint is usually the ORM patterns your team has accumulated.
Common questions.
Django + React or Django + HTMX?
HTMX for almost all internal tooling and admin. React only when you have genuine client-side state complexity. Most teams default to React out of habit and pay the bundle cost forever.
Should we use Django REST Framework or just async views?
DRF still excellent for full REST APIs with serialisers, permissions, and viewsets. Async views for narrow high-throughput endpoints.
PostgreSQL or MySQL for Django?
PostgreSQL for almost everything new. The full-text search, JSONB, and constraint support make a real difference once you scale past prototype.
Can you migrate us from old Django (3.x) to current?
Yes. Standard work. We do this with feature flags between major Django versions, not big-bang.
Ready to scope it?
A 25-minute call. We will tell you what we would do, what we would not, and whether a pod is the right shape.
Or stay in the loop. One engineering teardown a week.