Client & context
A startup wanted to build a DIY online product platform where users could customize products using images and templates. The system needed to:
- Handle computationally heavy image processing.
- Be ready for high load from marketing campaigns.
- Run on on-premises infrastructure for cost and control reasons.
I was responsible for designing and implementing both the application and the infrastructure.
Challenges
- Image processing algorithms were CPU-intensive and needed to be fast enough for real-time or near-real-time interactions.
- The solution had to run on self-hosted infrastructure, not a fully managed cloud service.
- The platform needed to scale horizontally, handle spikes, and protect against DDoS.
- The system had to support 18 languages and provide a responsive, fast UI that passes PageSpeed checks.
What I did
1. Designed and implemented custom image algorithms
- Created a set of custom algorithms for image processing and building DIY product previews.
- Tuned algorithms for performance while maintaining output quality.
2. Built high-load-ready on-prem infrastructure
- Designed and deployed an on-premises Kubernetes cluster with:
- Custom Docker images for the application and services.
- Multi-server architecture with load balancing.
- Pod scheduling tuned for heavy image processing workloads.
- Implemented CI/CD via a self-hosted GitLab instance, automating build and deployment pipelines.
- Added DDoS protection and resilience considerations to the infrastructure design.
3. Used Rust to accelerate PHP
- Identified critical performance hotspots in the PHP code.
- Rewrote those parts as a Rust-based PHP extension, dramatically improving performance where it mattered most.
4. Product, content, and UX work
- Led or coordinated content creation:
- Illustrations showing how the application works.
- Explanatory articles and documentation for users.
- Built a multilingual website, blog, and gallery (18 languages supported).
- Optimized the site to be fast enough to pass Google PageSpeed assessments.
Results
- The startup obtained a production-ready MVP capable of handling high load on self-hosted infrastructure.
- Critical image processing paths were significantly faster thanks to Rust acceleration.
- The platform combined robust infrastructure (Kubernetes, CI/CD, DDoS protection) with a fast, multilingual front end.
- This provided a strong foundation for further product development, fundraising, or customer acquisition.
Tech & responsibilities
- Role: Full-stack engineer, infrastructure architect, and product collaborator
- Technologies: PHP, Rust (PHP extension), Kubernetes, Docker, GitLab CI/CD, image processing, multilingual site, PageSpeed optimization
- Scope: Algorithms, backend, infrastructure, performance, content coordination, and UI performance
If you need a high-load-ready MVP with serious performance and infrastructure considerations from day one, I can help you design and implement a robust foundation rather than a throwaway prototype.
Back to all case studies
