Distributed Messaging System
Consolidated three separate messaging stacks into one shared platform, cutting infra costs by 99% across PadiUMKM.
99% infra cost cut across 3 PadiUMKM products
Yogi Prasetya
Software Developer based in Jakarta, Indonesia. Specializes in React, Next.js, TypeScript, Node.js, and the MERN stack. Builds fast, cost-efficient web systems for marketplace, messaging, POS, metaverse, and immersive-technology products.
Connect on LinkedIn (https://www.linkedin.com/in/yogiprasetya), GitHub (https://github.com/yogiprsetya), or Medium (https://medium.com/@yogiprsetya). Email: [email protected].
Software Developer - Jakarta, Indonesia
Software Developer based in Jakarta, Indonesia building fast, cost-efficient web systems with React, Next.js, and the MERN stack.
Full-stack developer with a frontend focus, building React libraries and products.
Consolidated three separate messaging stacks into one shared platform, cutting infra costs by 99% across PadiUMKM.
99% infra cost cut across 3 PadiUMKM products
Component library for Smarteye that gave every product team shared primitives and shortened design-to-ship time.
Shorter ship cycles, one UI system for all teams
Takes a job posting and rewrites your resume to match it, cutting time spent on each application.
Less time tailoring resumes to each job posting
Imports an OpenAPI spec or takes hand-added endpoints, giving both the same request builder, with no backend required.
Test APIs with or without OpenAPI docs

How consolidating fragmented features into a shared platform improves consistency, reduces maintenance, and accelerates delivery across multiple products.

A deep dive into rebuilding a chat system with a distributed messaging architecture that slashed infrastructure costs while keeping performance intact.

Lessons learned building a Web3 metaverse platform, from 3D experiences and wallet integrations to the trade-offs of early-stage immersive tech.

Practical prompt engineering techniques borrowed from Google’s approach to getting clearer, more useful outputs from large language models.