Roki Miftah Kamaludin

SUMMARY

Software Engineer with 10+ years of engineering experience, bridging hardware-software integration (SCADA, automation) and delivering modern end-to-end web applications. High-agency owner with a track record of shipping scalable products, modernizing legacy web codebases, and collaborating cross-functionally from system design to deployment.

EXPERIENCE

Full Stack Engineer

Self-Employed

January 2023 - Present, Remote

Shipped 5+ scalable web apps end-to-end with 90+ Lighthouse scores and responsive design.
Integrated third-party REST APIs with robust error handling, reducing parsing failures by 40%.
Debugged and modernized legacy JavaScript codebases, resolving 20+ critical issues.
Deployed and operated production apps on Linux VPS (Docker, reverse proxy), with CI/CD automation via GitHub Actions for repeatable builds and releases.
Collaborated asynchronously through GitHub PRs, code reviews, and daily updates to deliver milestones and maintain code quality.

Product Engineer

PT Len Industri (Persero)

November 2019 - June 2022, Bandung, Indonesia

Led the integration of technical components for 5+ complex engineering projects, ensuring strict compliance with design specifications.
Authored and maintained 30+ technical documents (system requirements & manuals), streamlining customer support processes and reducing tickets by 20%.
Fostered collaboration across a 15-person development team and external stakeholders, translating user feedback into 10+ actionable product improvements per year.
Facilitated continuous improvement initiatives that streamlined troubleshooting, reducing project delivery time by 15%.

Control Engineer

PT Len Industri (Persero)

October 2015 - October 2019, Bandung, Indonesia

Programmed and deployed SCADA systems for 3 major infrastructure projects, ensuring 99.9% system uptime for critical monitoring operations.
Conducted 20+ Factory Acceptance Tests (FAT) directly with clients to ensure systems met operational specifications.
Provided on-site technical support during installation, resolving critical hardware-software integration issues in real time to keep deployments on schedule.
Collaborated with 10+ external contractors to align technical requirements, reducing on-field discrepancies by 30% through preventative communication.

PROJECT

SAH-in Aja!

github.com/rokimiftah/SAH-in-Aja·sahin.biz.id

Architected an 8-module AI platform to support halal certification preparation, featuring Vision AI kitchen assessment, an OCR-based ingredient scanner detecting 50+ critical substances (E-codes, animal derivatives), a real-time Voice Audit Simulator, and auto-generated SJPH documents (7 templates).
Built a serverless real-time backend on Convex with 15+ tables, daily credit system with cron-based reset, multi-auth (Google OAuth + Magic Link), and freemium monetization supporting 6 feature-specific usage limits.
Impact: Targets Indonesia's 30M+ MSME market by enabling self-assessment of halal readiness prior to BPJPH audit; aims to reduce reliance on paid consultations (typically Rp 2-10M) and shorten preparation cycles (often 1-3 months) through AI-assisted pre-screening.

What The Pack

github.com/rokimiftah/whatthepack.today·whatthepack.today

Architected a multi-tenant AI Mission Control platform for D2C social commerce, implementing a zero-trust security model with Auth0 Organizations, RBAC (3 roles), and organization metadata to isolate and protect owner API keys—allowing low-privilege packers to trigger privileged workflows (e.g., ShipEngine label purchase) without direct credential access.
Engineered 6 specialized AI agents, including: Role-Aware RAG (role-based filtering before LLM calls), Voice Packing Assistant (Vapi.ai with 7 intents for hands-free warehouse operations), Chat Extraction (structured order form extraction from customer conversations), and Daily Briefing (LLM-generated analytics with trend signals).
Impact: Reduces operational risk in delegated workflows by enforcing credential isolation and least-privilege access, while streamlining routine logistics tasks through AI-assisted automation.

Travel Scam Alert

github.com/rokimiftah/scam.web.id·scam.web.id

Engineered an automated Reddit content pipeline across 5 travel subreddits using 18 scam-related keywords; processed posts/comments via an LLM to extract location signals, classify incidents into 15 scam types (e.g., taxi, romance, pickpocket), and generate warning summaries with prevention tips using a 0.6+ confidence threshold.
Built an interactive 3D globe visualization (react-globe.gl / Three.js) showing country-level aggregated reports with risk scoring; integrated a Vapi.ai voice assistant that queries the database via webhook tool calling, including country-alias mapping for 40+ naming variations.
Impact: Produces a searchable, geo-visualized dataset of crowdsourced scam reports spanning 50+ countries, enabling users to explore scam patterns and retrieve AI-generated safety summaries and voice-based briefings.

Mengenang Pahlawan

github.com/rokimiftah/mengenang-pahlawan·mengenangpahlawan.web.id

Developed an educational platform on Indonesian national heroes featuring LLM-powered persona chat (roleplay-style conversations), AI-generated quizzes with 5 question types (birth year, birthplace, era, recognition basis, biography highlights), and student-friendly content simplification using OpenAI.
Built a real-time serverless backend on Convex with 7 database tables, a gamification system (daily quiz caps, point accumulation with perfect-score bonuses, practice-mode detection), multi-auth (Google OAuth + email OTP via Resend), and era-based filtering across 5 historical periods.
Impact: Converts structured hero biographies into interactive learning experiences (chat + quizzes), supporting exploration of a catalog of 200+ national heroes through AI-assisted summaries and assessment flows.

GitHub Issue Watcher

github.com/rokimiftah/github-issue-watcher·giw.web.id

Engineered a distributed, LLM-assisted GitHub issue analysis pipeline on Convex (serverless), including paginated GraphQL fetching (up to 4,000 issues/repo), a priority task queue with mutex locking, token-bucket rate limiting (20 RPM), and automated API key rotation with 24-hour expiry management via cron jobs.
Built an end-to-end relevance scoring workflow using weighted signals (Title 0.45, Body 0.35, Labels 0.20) with DeepSeek/Qwen LLM, delivering results via a React/Mantine dashboard with progressive email notifications (partial + final), multi-auth (GitHub OAuth + email OTP), and 1-hour caching to reduce redundant API calls.
Impact: Provides keyword-based discovery and AI-generated relevance rationales for large repositories, helping teams prioritize candidate issues more efficiently while preserving traceable evidence for each match.

EDUCATION

Associate Degree in Electrical Engineering

Jenderal Achmad Yani University · Cimahi, Indonesia

SKILLS

Core Stack: Next.js, TypeScript, React, Tailwind CSS, Node.js.

Languages: TypeScript, JavaScript (ES6+), SQL, Python, HTML5, CSS3.

Backend & Data: REST APIs, Convex (Backend-as-a-Service), PostgreSQL, Auth (OAuth/Magic Link), Vector DB/Embeddings.

DevOps: Docker, Linux VPS, Caddy (Reverse Proxy/Web Server), GitHub Actions (CI/CD), Self-Hosting.

AI: LLM Integration (OpenAI/Local), RAG, Web Scraping (Cheerio/Puppeteer), Data Parsing.

Tools: Git/GitHub, Postman, VS Code, npm, Biome/Prettier.

Core Competencies: End-to-End Delivery, System Design, Ownership (High Agency), Root-Cause Analysis, Problem Solving, Code Quality, Scalability, Performance Optimization, Legacy Refactoring, Technical Documentation, Cross-Functional Collaboration, Teamwork.