Kaka
Alfuna
Zhafran.
Fullstack & DevOps engineer — systems, cloud, and AI/ML in production.
−7.36, 109.90
Who I am,
where I build.
I architect scalable systems, optimize cloud infrastructure, and ship AI/ML pipelines — from multi-region ticketing at peak concurrency to zero-downtime bank campaigns serving 100k+ users.
Kaka Alfuna Zhafran · Fullstack & DevOps · Cloud · AI/ML
One engineer,
four surfaces,
one product.
I own the whole system — from the first pixel a user sees to the container running in production at 3am.
Frontend
Interfaces built for speed and discoverability.
Career arc, told in chapters.
Swipe the timeline — each card is one chapter.
Each year added a layer to the stack. Hover a node to read its chapter.
Eduwork — Full Stack Intern
Built Laravel + Vue apps end-to-end — schema design, REST APIs, and production delivery through the Eduwork acceleration program.
Cakra Techno — Frontend & Automation
Shipped Next.js and Vue interfaces integrated with ML APIs. Automated E2E and load testing with Selenium and Locust — cut manual QA cycles by 40%.
Five Elements — Global platforms
Engineered ticketing and social infra across Australia, Singapore, Malaysia, and Indonesia. Solved ticket-war race conditions with Redis locks, then MongoDB distributed transactions at 99.9% consistency.
AI products in production
Delivered Kisum (predictive event analytics) and Carousel Studio (programmatic content SaaS) — hybrid CatBoost + LLM pipelines and cost-efficient DOM-to-image generation at scale.
Systemcraft — Lead Engineer
Lead fullstack & DevOps for enterprise bank campaigns — 100k+ users, 1k concurrent peaks, AWS ECS auto-scaling, 100% availability, and GEO-driven visibility in modern AI search engines.
01 / 05
Bank campaign infra at national scale.
Spearheaded high-stakes promotional web apps for a major commercial bank — 100,000+ total users with up to 1,000 concurrent actives. Architected AWS ECS auto-scaling for zero downtime during traffic spikes.
Active users handled with elastic ECS auto-scaling and sub-second page loads.
Availability during campaign spikes
Total campaign users served
End-to-end products, shipped solo or with small teams.
Kisum
Event intelligence & analytics
Carousel Studio
Data-backed decisions for promoters — before showtime.