Vittorio Esposito

Platforms like Vercel, Netlify, Render handle 90% of infra headaches out of the box: scaling, HTTPS, CDNs.

Trade-offs:

• Slightly higher cost at scale

• Less control over low-level configs
  

  But for an MVP, the gain in speed and simplicity outweighs everything else.
  Optimize for learning, not infra purity.

At one client, data-heavy endpoints were causing latency and high infra costs.
Instead of scaling servers, we pre-computed results nightly and stored them on S3.
Impact: faster API, lower costs, happier users.
Simple pipelines beat complex real-time systems when real-time isn’t necessary.

On a project, builds took ~25 minutes. We tweaked CI to cache dependencies properly and parallelize steps.
Builds dropped to ~7 minutes.
That’s 18 minutes saved × dozens of commits per week = hours of team time back.
Lesson: optimize dev workflow, not just product features.

Most stable MVPs I’ve seen were built with boring, proven stacks: React, Node, Postgres.

Why?

• Easier to hire for

• Well-documented

• Tons of community support
  

  Cutting-edge stacks are tempting, but often add hidden risks and time.

One client thought they needed bigger servers.
We ran profiling first: turns out a single query was dragging everything down.
Fix: added indexing + caching.
Result: 3× faster API, no new servers needed.
Lesson: scaling starts with understanding, not throwing money at infra.

For consulting projects in big enterprises, Docker is mandatory: controlled environments, CI/CD pipelines, reproducibility.

For scrappy MVPs, I often skip it at first. Why?

• Local dev simpler

• Hosting easier with managed services

• Fewer moving parts for founders to learn

  
  Docker becomes useful once you have a team or complex infra, not before.

Next.js + a simple backend covers 80% of MVP needs: landing page, auth, dashboards, payments.
Even if the long-term product will be mobile or IoT, a Next.js prototype validates the concept faster and cheaper.
You can always rebuild in native tech once the idea proves itself.

We baked screen reader testing into every story.
Unexpected benefit: designs became clearer and more consistent even for sighted users.
Accessibility isn’t just compliance, it’s good UX for everyone.

One client had a dev team waiting 20–30 minutes for builds to finish.
We moved to Vercel + incremental static regeneration.
Result: faster previews, faster releases, happier stakeholders.
Lesson: shaving time off builds compounds into weeks saved over a year.

Postgres on RDS, Supabase, or Firebase > self-managed DB for early products.
You save:

• Setup time

• Maintenance headaches

• Security patches
  

  Yes, you’ll pay a bit more monthly, but the trade-off in speed and peace of mind is worth it.