labs.alvarny.studio · v3.21.0

Web apps,
built to ship
and scale with you.

A two-person engineering team for founders and ops leaders who need real software — dashboards, internal tools, customer portals, B2B SaaS — typed end-to-end, deployed on day one, owned by you.

./book-intake.sh → cat stack.json
8–12wkto launchable v1
99.95%SLA on hosted apps
0dependencies on us, post-launch
~/labs/intake — fish [zsh]
labs ➜ scope --from idea --to v1
parsing idea.md...
found 3 user types, 7 routes, 1 admin surface
estimated effort: 9.5 weeks · 2 engineers
labs ➜ build --stack ts/next/postgres
✓ types end-to-end (zero any)
✓ migrations versioned, rollbackable
✓ auth, billing, audit log: included
✓ CI/CD, preview envs per PR
labs ➜ handover --to client
→ repository: github.com/yourco/yourapp
→ infra: vercel + neon (you own both)
→ docs: runbook.md, adr/, decisions/
labs ➜
// 02 — what we build

Three flavors of app work.

001 / b2b saas

Founder MVPs & v1s

You have customers who'll pay. We turn the deck and the spreadsheet into a real, multi-tenant product with auth, billing, and dashboards from day one.

  • Multi-tenant from the schema up
  • Stripe billing + usage metering
  • Admin console + impersonation
002 / internal tools

Ops & back-office apps

Replace the Airtable + Zapier + spreadsheet sprawl your team has outgrown. Custom CRUD, workflows, approvals — fast to build, easy to maintain.

  • SSO & role-based access
  • Audit log everything
  • Bulk import/export, CSV-friendly
003 / customer portals

Client & partner portals

The "logged-in version" of your business — invoices, files, project status, support, account settings. Plays nicely with your existing stack.

  • Embeds into your marketing site
  • API-first; we don't lock you in
  • White-label friendly
// 03 — the stack

Boring, typed, and hireable.

Frontendbrowser layer
Next.js 15 React 18 TypeScript Tailwind shadcn/ui TanStack Query
APIedge / server
tRPC Hono Zod Auth.js Inngest (jobs)
Datapersistence
Postgres Drizzle ORM Neon / Supabase Redis S3 / R2
Infradeploy & observe
Vercel Cloudflare Sentry PostHog GitHub Actions
Integrations3rd-party
Stripe Resend Twilio OpenAI / Anthropic WorkOS / Clerk
// 04 — recent work

Plate.io — restaurant ops in one tab.

A multi-location restaurant group came to us with seven SaaS subscriptions, three spreadsheets, and one frustrated COO. We replaced four of those with a single internal app — staff scheduling, inventory, vendor invoicing, and shift handover — built on Postgres, Next.js, and a clean API the team's now extending themselves.

11wk kickoff to v1 launch 3 modules shipped, 2 engineers
−€38k/yr SaaS spend killed replaced 4 vendor subscriptions
99.97% uptime, last 6mo single Sev-2 incident, 9min
12kreq/d peak API throughput p95 latency: 180ms
// 05 — engineering doctrine

How we actually work.

tenet 01

Types are the docs.

End-to-end TypeScript, Zod-validated boundaries, generated DB types. If it compiles, the contract holds. No hand-rolled JSON shapes hiding in API routes.

tenet 02

Migrations or it didn't happen.

Every schema change is a versioned, reviewed, rollbackable migration. No "I just ran this on prod" — Drizzle migrations, checked into your repo from week one.

tenet 03

You own the keys.

Vercel account in your name. Database in your name. Stripe, GitHub, DNS — your name. We're collaborators, not landlords. Day-one handover is the default.

tenet 04

Boring tech where it counts.

Postgres, not the database-of-the-month. React, not the framework-of-the-week. We pick novelty deliberately and only when it earns its keep — and we write down why.

// 06 — who you'll work with

You're talking to the engineers, not a sales rep.

Y

Y. Sahl

PRINCIPAL · FULL-STACK · BERLIN
github.com/yrsahl ~/cv.pdf

Eight years shipping production TypeScript, four of those leading a small team at a Series-B SaaS. Architecture, frontend, infra, on-call. Writes the PRs and reviews them.

M

M. Reiter

ENGINEER · BACKEND · REMOTE / EU
github.com/mreiter ~/work.md

Postgres, distributed systems, payments. Spent five years inside a fintech learning what breaks at scale, so your app doesn't have to learn the same lessons in production.

// 07 — pricing

Three engagement shapes.

SPRINT
€9.8k/ 4 wks

A focused sprint to ship one well-defined module — a portal, a dashboard, an integration, a migration off something painful.

  • 1 engineer, 1 module
  • Daily Slack updates
  • You own the repo from day one
  • 30 days post-sprint support
Start a sprint →
V1
€38–58k/ 8–12 wks

A launchable product. Auth, billing, dashboard, admin, deploy, monitoring. The full path from idea to first ten paying users.

  • 2 engineers, full stack
  • Weekly demos, preview envs per PR
  • Production-ready on launch day
  • 90 days warranty + handover
Scope a v1 →
EMBED
€16k/ month

For post-launch product work. We embed alongside your team for a fixed monthly fee — features, bug-bash, migrations, incident response.

  • 1.5 engineers, ~60h/wk
  • Shared on-call rotation (optional)
  • Roadmap planning included
  • 30 days notice, either side
Talk to us →
// 08 — intake

Tell us what you're building.

Project intake /dev.

Five fields. We reply within 4 working hours with either a 30-minute call invite, or an honest "this isn't a fit, here's who's better."

$ what happens next
1. We read it (a human, same day).
2. 30-min intake call within the week.
3. Fixed-fee proposal in 48h, or a no.
// 09 — faq

Things engineers ask first.

Why TypeScript / Next / Postgres? Why not [framework-of-the-week]?
Because they're boring, hireable, and well-documented. Three years from now you'll be able to find a contractor or a hire who knows them, on three continents. We've shipped on Phoenix, Rails, Go, and Rust too — happy to use what fits, but the default is what your future self can maintain.
Do we own the code and the infrastructure?
Yes, completely. The GitHub repo is in your org, the Vercel/Neon/Stripe accounts are in your name, and we're collaborators. We've never held a client's keys hostage and we never will. Day-one handover is the default.
Can we start with a small sprint before committing to a v1?
Yes — that's literally the SPRINT tier. €9.8k, 4 weeks, one focused module. Most clients use it as a working trial. If it goes well we roll into a v1 with the sprint fee credited.
What about AI features?
We've shipped LLM-backed features into production apps — RAG over docs, classification, agents, structured extraction. We're skeptical of AI-for-AI's-sake but we know how to make it boring enough for prod (eval harnesses, fallbacks, cost controls).
Do you do mobile (iOS / Android / React Native)?
Not natively. We default to high-quality responsive web + PWA, which covers 90% of "we need an app" requests. For genuine native apps we partner with a small studio in Lisbon — happy to make the intro.
What's the smallest project you'll take?
A 2-week sprint, around €5k, for something tightly scoped (a single integration, a migration, a polish pass). Below that we're not the right fit — you want a freelancer, and we know good ones.