Changelog
Platform Evolution
Every feature, improvement, and milestone shipped across CorePlexML — from day one to today.
v2.6February 2026Latest
ML Studio & Developer Experience
- ML Studio What-If Analysis — interactive scenario comparison with SHAP contributions and PDP plots
- SweetAlert2 dark mode integration across all UI dialogs and confirmations
- Sphinx-powered API documentation site with full endpoint coverage
- Global search palette (Ctrl+K) for instant navigation across all platform resources
- Onboarding tour for first-time users with guided walkthrough
v2.5February 2026
Conversational Dataset Builder
- Complete redesign of Dataset Builder with LangGraph + LLM-driven conversation
- Natural language data preparation — describe transformations in plain English
- Support for OpenAI (GPT-4o-mini) and Anthropic (Claude) as LLM backends
- Multi-step pipeline: upload, profile, clean, engineer features, transform, export
- Transparent script execution — every transformation shown as executable Python code
v2.4February 2026
Advanced Deployment Strategies
- A/B testing with Bayesian and Frequentist statistical analysis
- Canary deployments with progressive traffic ramp and auto-rollback
- Blue-green deployments with instant swap and zero-downtime rollback
- Shadow deployments for passive model validation on production traffic
- Drift detection with Population Stability Index (PSI) monitoring
- Deployment health checks with latency percentile tracking (p50, p95, p99)
v2.3February 2026
Privacy Suite & SynthGen
- Privacy Suite with 72+ PII type detection (names, emails, SSN, medical records, etc.)
- Four compliance profiles: HIPAA, GDPR, PCI-DSS, CCPA
- Six transformation actions: mask, redact, hash, encrypt, generalize, suppress
- SynthGen synthetic data generation with CTGAN, CopulaGAN, and TVAE engines
- Quality metrics for synthetic data: statistical similarity, privacy score, utility score
- PDF compliance reports with detection summaries and transformation audit trails
v2.2January 2026
MLOps Model Registry & Alerts
- Model registry with semantic versioning and stage management (dev → staging → prod → archived)
- Model cards with documentation, metrics, and lineage tracking
- Multi-channel alert system: Slack, email, and webhook notifications
- Six configurable alert metrics: drift PSI, accuracy degradation, error rate, latency, staleness, anomaly
- Alert suppression windows and escalation rules
- Auto-retraining policies: scheduled (cron), drift-triggered, and performance-based
v2.1January 2026
Batch Predictions & Reports
- Batch prediction jobs with CSV upload and async processing
- WebSocket streaming for real-time batch inference progress
- Seven report types: model performance, feature importance, drift analysis, deployment summary, and more
- PDF report generation with automated charts and visualizations
- Jupyter notebook export for any experiment — reproduce results locally
v2.0January 2026
Multi-Tenant Enterprise Platform
- Multi-tenant architecture with project-level isolation
- Role-Based Access Control (RBAC): owner, editor, viewer per project
- SSO/SAML integration for enterprise identity providers
- OAuth support for Google and GitHub
- Stripe billing with Free and Pro subscription tiers
- Usage metering and quota enforcement
- Admin dashboard for user management and platform settings
v1.0December 2025
AutoML Foundation
- AutoML with H2O.ai — GBM, XGBoost, Deep Learning, GLM, and Stacked Ensembles
- Automated feature engineering and hyperparameter optimization
- Dataset management with version tracking and column metadata
- Experiment leaderboard with model comparison metrics
- One-click model deployment to staging and production
- REST API for real-time inference
- Project-based organization with PostgreSQL persistence