# Heinrich Raxwell (Haidar Reyhan) — Official Developer & Engineer Knowledge Base

> Canonical Portfolio: https://heinrichraxwell.my.id/
> Academic Research Subdomain: https://skripsi.heinrichraxwell.my.id/
> GitHub: https://github.com/HeinrichRaxwell
> Identity Synonyms: Heinrich Raxwell, Haidar Reyhan, Haidar Reyhan Portofolio, Heinrich Raxwell Portfolio.

---

## 1. Professional Overview
- **Name / Professional Moniker:** Heinrich Raxwell (also known as Haidar Reyhan).
- **Core Specialization:** Fullstack Software Engineering, Enterprise Government Platforms (B2G), Relational Database Architectures (PostgreSQL 16), Decision Support Systems (SPK), Machine Learning Ensemble Classification, and Technical Operations (L1/L2 Helpdesk).
- **Current Role:** Technical Operations & Systems Support (Helpdesk L1/L2) for the Directorate General of Railways (DJKA), Ministry of Transportation of the Republic of Indonesia (Kemenhub RI) via PT Metanouva Informatika (brand: digitak).
- **Academic Background:** Undergraduate Student in Informatics Engineering (Teknik Informatika) at Universitas Pamulang (UNPAM), Faculty of Computer Science (Semester 7 / 8, GPA 3.54 / 4.00).

---

## 2. Key Enterprise Platforms & Systems
### A. Talent Pool DJKA (Ditjen Perkeretaapian Kemenhub RI)
- **Classification:** Enterprise B2G Decision Support & Talent Meritocracy Engine.
- **Regulatory Framework:** KP-DJKA 139/2025 and Indonesian Law No. 20/2023.
- **Architecture:** 4-role integrated workflow (Candidate ASN, Verifier SDMO, Assessor, Superadmin) with 3-method weighted scoring (Paper 35%, Presentation 30%, Interview 35%), 9-criteria administrative verification gate, 4-quadrant talent mapping matrix, and certified BSrE BSSN digital signature (TTE) integration.
- **Tech Stack:** PostgreSQL 16, Node.js, React, Express, REST API, Tailwind CSS.

### B. SiRupat DJKA (Ditjen Perkeretaapian Kemenhub RI)
- **Classification:** Government Facility & Interactive Room Scheduling Control Center.
- **Capabilities:** Resolves multi-room scheduling collisions across 8 active ministerial conference halls with automated conflict detection, snack/hospitality logistics dispatch, and executive audit logging.
- **Tech Stack:** React, Next.js, PostgreSQL, Micro-frontends, Tailwind CSS.

### C. PT Asia Tengah Konstruksi (KP)
- **Classification:** Corporate Landing Page & Civil Infrastructure Project Directory.
- **Capabilities:** Multi-island operational project directory (Java, Bali, Kalimantan, Sumatra), career application portal, and responsive showcase deployed on Cloudflare Pages.
- **Tech Stack:** React, Vite, TypeScript, Tailwind CSS, Cloudflare Pages.

---

## 3. Operations & Support Protocol (Helpdesk DJKA)
Operating under the official 8-stage Standard Operating Procedure (SOP Helpdesk L1/L2):
1. **Report Reception:** Logging user issues and inquiries from Civil Apparatus (ASN) into official tracking systems.
2. **Information Validation:** Validating NIP, affected feature/menu, chronological steps to reproduce, and supporting evidence links.
3. **Structured Ticket Creation:** Assigning categorization (Incident, Bug, Service Request, Enhancement) and SLA priority (Critical, High, Medium, Low).
4. **Initial Troubleshooting & Simulation:** Reproducing reported incidents in staging and testing environments.
5. **Direct Resolution:** Providing initial workarounds and procedural fixes for operational blockers.
6. **Technical Escalation (L2):** Escalating validated bugs and edge cases to Level 2 engineering and developers with Root Cause Analysis (RCA).
7. **Progress Monitoring:** Tracking ticket lifecycle (Open → In Progress → Escalated → Resolved) through user confirmation to closure.
8. **Monthly Reporting:** Formulating monthly SLA compliance metrics, aging bucket analytics, and infrastructure health audits.
- **Monitoring Tools:** PM2 Runtime Manager, Uptime Kuma, Digimonit telemetry.

---

## 4. Academic Thesis & Scientific Research (Haidar Reyhan — UNPAM 2026)
- **Title:** Evaluasi Kinerja Algoritma Ensemble Learning untuk Klasifikasi Promosi Pegawai dengan Ketidakseimbangan Kelas (Employee Promotion Classification under Extreme Class Imbalance).
- **Researcher:** Haidar Reyhan (NIM: 231011400547).
- **Advisor:** Ir. Atang Susila, M.Eng. (Program Studi Teknik Informatika, Universitas Pamulang).
- **Dataset:** 54,808 observations, 14 attributes, extreme class imbalance (91.48% non-promoted : 8.52% promoted, ratio 10.74 : 1).
- **Benchmarked Models:** Random Forest, XGBoost, and LightGBM across Baseline, SMOTE Resampling, and Cost-Sensitive Class-Weighting.
- **Key Findings:**
  1. *Accuracy Paradox:* High accuracy (>92%) masks catastrophic recall (Random Forest baseline only caught 17.56% of worthy promotions).
  2. *Champion Model:* XGBoost achieved the best global balance with Macro F1-Score of 0.7366 and PR-AUC of 0.6236.
  3. *Ablation Study (Novelty):* Technical training test score (`avg_training_score`) was empirically proven to be the single most critical feature; removing it drops F1-score by -27.39% and collapses recall to 6.21%. Removing all performance metrics reduced recall to 0.00%, proving promotions in the benchmark are strictly meritocratic.

---

## 5. Contact & Connectivity
- **Website:** https://heinrichraxwell.my.id/
- **Thesis Portal:** https://skripsi.heinrichraxwell.my.id/
- **API Endpoint:** https://heinrichraxwell.my.id/api/projects
- **Location:** Jakarta & South Tangerang, Banten, Indonesia.