DEV Community

Nadim Chowdhury
Nadim Chowdhury

Posted on

How I Built an 80-Tool Web Platform with $0 Monthly Cloud Compute Costs

When building side projects or indie web applications, one of the biggest risks is infrastructure creep.

You start with a simple idea, add a PostgreSQL database, spin up an AWS ECS container for background processing, set up Redis for caching, and suddenly you are paying $80/month in cloud infrastructure before acquiring your first 100 users.

A few weeks ago, I challenged myself to build Omnikite — a massive web platform with 80 high-utility tools (PDF processing, cryptography, formatters, image minifiers, financial matrices) — with $0 in ongoing backend compute costs.

Here is how the architecture works and what I learned.

1. The Strategy: Shift 100% of Compute to the Client

Traditional utility websites process files on their own servers:

  • User uploads a 10MB PDF $\rightarrow$ Server downloads $\rightarrow$ Server runs Python script $\rightarrow$ Server uploads to S3 $\rightarrow$ User downloads result.

This architecture has three fatal flaws for an indie founder:

  1. Server CPU costs scale with traffic.
  2. Bandwidth egress costs eat into your margins.
  3. Handling user files creates severe GDPR/privacy liability.

Instead, we built every single utility to run directly in the user's browser using:

  • WebAssembly (WASM) for binary parsing and PDF manipulation.
  • Web Crypto Subtle API for SHA-256, HMAC, and PBKDF2 hashing.
  • Canvas 2D API for image compression and format conversions.
  • Pure JavaScript / TypeScript state for real-time mathematical modeling.

2. Zero-Database Architecture with Static Generation

Omnikite uses Next.js 16 (App Router) with Static Site Generation (SSG).

All 99 routes (80 tool pages + 12 category directories + home + tools directory) are pre-rendered into static HTML at build time.

  • Hosting: Deployed on Vercel's global Edge CDN.
  • Time to First Byte (TTFB): Under 25ms worldwide.
  • Database Maintenance: None. User favorites and recents are stored locally in the user’s browser using useSyncExternalStore and localStorage.

3. SEO-First Schema Engineering

To compete with legacy utility websites on Google Search, we implemented a 4-layer structured data schema across every page:

  1. WebApplication Schema: Informs Googlebot that the page is an interactive web tool.
  2. BreadcrumbList Schema: Generates clean search snippet hierarchy.
  3. FAQPage Schema: Qualifies the page for Google rich result FAQ dropdowns.
  4. HowTo Schema: Step-by-step instructional cards that appear directly on Google search results.

4. The Results

  • Active Tools: 80 Flagship Utilities across 12 domains.
  • Performance: 100/100 Google Lighthouse across Desktop & Mobile.
  • Monthly Cloud Bill: $0.00.

If you want to explore the platform or see the UX in action, check it out here:
👉 https://omnikite.vercel.app

What side projects are you building right now? Have you considered moving parts of your backend into client-side Web Workers or WASM?

Top comments (0)