External and internal penetration testing with structured vulnerability reporting. OWASP Foundation member who ships security fixes to open source (OWASP, NVIDIA) and hardens every app with RBAC, JWT and zero-trust thinking.
How I test & report
- Scope & recon. Agree scope and rules of engagement, then map the attack surface — external or internal — before touching anything.
- Scan & enumerate. Find exposed services, endpoints and misconfigurations with tools such as Kali Linux and Burp Suite.
- Exploit & validate. Confirm each finding with a safe proof of concept and rate its real impact, without causing damage.
- Report. Write a structured vulnerability report: description, severity, reproduction steps, evidence and clear remediation advice.
- Fix & retest. Help close the gap — RBAC, JWT, CSP, zero-trust hardening — then re-test to verify the hole is really shut.
Toolkit
- Penetration Testing
- Kali Linux
- Burp Suite
- Zero-Trust Architecture
- Cloud Security
- OWASP
Projects
Bug bounty & reports
- TrickHLA — High Level Architecture Simulation Engine — NASA · Logic Bug
- MSTest Framework — Microsoft · Architecture Contract Failure
- Keyborg Accessibility Library — Microsoft · Memory Leak
- Infrastructure Controller — NVIDIA · Vulnerability
- Security Cheat Sheet Series — OWASP Foundation · Vulnerability Research
- Transformers Engine — Hugging Face · Logic Bug
- Site Kit Analytics — Google / WordPress · Logic Bug
- Agent Skills — Google Engineering Fellow · Configuration Error
- ADK Python & Samples — ADK Ecosystem · Silent Type Error
Computer vision with YOLOv8 and OpenCV, LLM products on Gemini, Groq and NVIDIA APIs, retrieval-augmented generation and ML classifiers — served through FastAPI and shipped to production.
From data to deployed model
- Collect & annotate. Gather and label data (Roboflow, Kaggle) so the model learns from clean, representative examples.
- Train & fine-tune. Fine-tune models such as YOLOv8 with PyTorch / Hugging Face and track what actually improves results.
- Evaluate. Measure with mAP, precision and recall — not vibes — and keep the evaluation reproducible.
- Serve. Expose the model through FastAPI behind a Next.js product UI, deployed on Railway / Vercel / Cloud Run.
- Secure & monitor. Add audit trails, JWT auth and zero-trust boundaries so AI features are safe to run in production.
Toolkit
- Claude Code
- AI Agents
- MCP
- RAG
- Google Gemini
- FastAPI
- YOLOv8
- OpenCV
- Hugging Face
Projects
Competitions
- Vibe Coding Winner — Google · June 2026
- Web Development Category Participant — Codelab Indonesia · June 2026
- Participant Data Automation Competition — PUMA Information System President University · June 2026
- AI Open Innovation Challenge 2026 — President University & Coordinating Ministry of Economic Affairs RI · February – August 2026
Full-stack systems end to end: Laravel, Next.js, Go and ASP.NET on PostgreSQL. An ERP/POS platform serving thousands of customers, 1.34 s load time and 91% good LCP, plus 3,300+ GitHub contributions.
From idea to production
- Design. Model the domain, pick the architecture (monolith, clean architecture or microservices) and design the database first.
- Build. Ship features end to end with Laravel, Next.js, Go or ASP.NET — typed, reviewed and documented.
- Test. Unit and integration tests plus QA / UAT so releases are boring in the best way.
- Ship. Containerise and deploy (Docker, Vercel, Railway) with environment-specific config and secrets handled safely.
- Optimise. Measure Core Web Vitals, cache aggressively and put a CDN in front — fast is a feature.
Toolkit
- Laravel
- Next.js
- TypeScript
- Golang
- PostgreSQL
- Astro
- WhatsApp Gateway
- Email Gateway
- Webhooks
Projects
Experience
- Software Developer Intern — PT GRAHABUANA CIKARANG · August 2026 - December 2026
- Full Stack Engineer — ASHAR GROSIR PARFUM BEKASI · December 2025 - Present
- Web Development Intern — GAOTEK INC · April 2026 - July 2026
Containerised, observable and fast: Docker, Redis, RabbitMQ and OpenTelemetry behind Cloudflare CDN, deployed on Vercel, Railway and Google Cloud Run. Completed AWS Cloud Practitioner Essentials and AWS Technical Essentials training.
The release pipeline
- Commit. Small, reviewable commits on short-lived branches keep the main branch always releasable.
- Build. Reproducible Docker builds so what runs on my laptop is what runs in production.
- Test. Automated tests gate every change before it gets near users.
- Security scan. Dependency and configuration checks catch known-vulnerable packages and unsafe defaults early.
- Deploy. Roll out to Vercel, Railway or Google Cloud Run behind Cloudflare, with an easy way back.
- Observe. Telemetry (OpenTelemetry, Prometheus-style metrics) shows how the release behaves in the real world.
Toolkit
- Docker
- Cloudflare
- Firewall & WAF
- Linux Server
- Vercel
- Railway
- Google Cloud Run
- AWS
Projects
Cloud training
- AWS Cloud Practitioner Essentials – Training Completion — Amazon Web Services (AWS) · March 2026
- AWS Technical Essentials — Amazon Web Services (AWS) · January 2026