SKKNI BNSP SUPERVISOR LEVEL•Kementerian Komunikasi dan Digital RI (Komdigi) — DTS• Requirements Completed (Pending Certificate Issuance)

Komdigi DTS: Data Scientist Supervisor

SKKNI BNSP Intermediate Machine Learning Pipelines, Model Validation & Technical Governance (20 JP)

KomdigiSKKNI BNSPModel ValidationFeature EngineeringCross-ValidationSupervisory Governance

TRACK COMPLETION PROGRESS:20 JP Completed (100%)

TIMELINE: BATCH 1 (2026) – 2026STATUS: REQUIREMENTS COMPLETED (PENDING CERTIFICATE ISSUANCE)

CORE OBJECTIVE & TECHNICAL FOCUS

"Mastered intermediate data science supervision, robust cross-validation architectures, high-performance ensemble modeling, and governance pipelines under the 20 JP Komdigi DTS Supervisor curriculum."

// ENROLLED MODULES & CURRICULUM (6)

UK 01: Memvalidasi Data & Data Quality AssuranceIntermediate (20 JP)

Multi-dimensional data quality auditing, schema drift detection, and statistical distribution consistency checks.

Status: CompletedCurriculum ↗

UK 02: Menentukan Objek Data & Problem FormulationIntermediate (20 JP)

Translating enterprise business objectives into measurable machine learning targets and technical constraints.

Status: CompletedCurriculum ↗

UK 03: Mengkonstruksi Data & Advanced Feature EngineeringIntermediate (20 JP)

Non-linear feature transformations, interaction terms, dimension reduction (PCA), and automated feature pipelines.

Status: CompletedCurriculum ↗

UK 04: Membangun Skenario Data & Validation Split ArchitecturesIntermediate (20 JP)

Stratified $k$-fold cross-validation, temporal train-test splits for time series, and data leakage prevention.

Status: CompletedCurriculum ↗

UK 05: Membangun Model & Hyperparameter OptimizationIntermediate (20 JP)

Training ensemble classifiers (Random Forest, Gradient Boosting, XGBoost) and Bayesian/GridSearchCV tuning.

Status: CompletedCurriculum ↗

UK 06: Mengevaluasi Hasil Pemodelan & Production DiagnosticsIntermediate (20 JP)

Comprehensive evaluation matrices (ROC-AUC, Precision-Recall curves, SHAP explainability, and error drift analysis).

Status: CompletedCurriculum ↗

NOTE

Supervisory Occupational Scheme Overview:

- Authority: Kementerian Komunikasi dan Digital RI (Komdigi) — Digital Talent Academy

- Occupational Standard: Supervisor Ilmuwan Data (Data Scientist Supervisor) — SKKNI BNSP

- Duration: 20 Jam Pelajaran (JP) Intensive Self-Paced Track

- Official Program Portal: `s.komdigi.go.id/data-scientist-supervisor`

- Status: 100% Modules & Final Assessment Completed (Awaiting certificate release)

01. Program Scope & Supervisory Focus

The Data Scientist Supervisor program is an intermediate professional development track designed for senior practitioners overseeing end-to-end machine learning lifecycle management.

Key areas of focus include data quality governance, advanced feature construction, rigorous validation design without target leakage, multi-model hyperparameter optimization, and post-training explainability.

02. Six Supervisory Unit Competencies

Unit Dimension	Competency Standard	Applied Technical Deliverables
01. Data Validation	UK 01: Memvalidasi Data	Schema integrity checks, distribution drift tests, and automated data assertion testing.
02. Objective Scoping	UK 02: Menentukan Objek Data	Metric translation (ROI vs F1-score), risk tolerance framing, and KPI definition.
03. Feature Engineering	UK 03: Mengkonstruksi Data	Custom Scikit-Learn transformers, encoding high-cardinality features, and interaction modeling.
04. Validation Architecture	UK 04: Membangun Skenario	5-Fold Stratified Cross-Validation pipelines preventing data leakage across train-test boundaries.
05. Ensemble Modeling	UK 05: Membangun Model	Multi-model benchmarking (Random Forest, LightGBM, XGBoost) with RandomizedSearchCV.
06. Model Diagnostics	UK 06: Evaluasi Pemodelan	Precision-Recall trade-off optimization, cost-benefit matrix modeling, and model explainability.

03. Validation & Governance Flow

MERMAID
8 LINES

flowchart TD
    A["Raw Feature Space"] --> B["Quality Validation & Preprocessing"]
    B --> C["Stratified 5-Fold Cross-Validation Architecture"]
    C --> D1["Ensemble: Random Forest"]
    C --> D2["Ensemble: Gradient Boosting"]
    C --> D3["Baseline: Logistic Regression"]
    D1 & D2 & D3 --> E["Multi-Metric Evaluation (AUC / F1 / Latency)"]
    E --> F["Champion Model Selection & Supervisory Governance"]

IMPORTANT

Operational Status: 20 JP coursework, hands-on modeling notebooks, and final assessments completed with zero defects.