I’m interested in machine learning, statistics, mathematics, and the software systems that turn data into useful applications. My projects explore time-series forecasting, equipment-health modeling, anomaly detection, and applied AI.
Outside of studying and coding, I enjoy Brazilian Jiu-Jitsu (BJJ), Muay Thai, watching UFC, and weightlifting (I am in the 1000lbs club!).
From my latest work to earlier projects:
| Project | Focus |
|---|---|
| Anomaly Analysis | An ongoing independent study of anomaly-detection calibration and LLM-assisted verification: precision–recall tradeoffs, decision thresholds, false-alarm analysis, and scenario-based evaluation. |
| Predictive Maintenance of Rotating Equipments | An internship proof of concept connecting sensor and maintenance data with temporal feature engineering, Random Forest regression, and remaining-useful-life modeling. Covers operating-cycle segmentation, near-failure error analysis, target validity, and presentation to executive and global engineering stakeholders. |
| Demand Forecasting | Multivariate shipment-demand forecasting with Temporal Fusion Transformers and PyTorch Forecasting: hierarchical features, leakage-aware feature engineering, temporal holdout validation, real-unit error metrics, and SARIMAX baseline context. |
| Predictive Maintenance of CNC Machine | Industrial telemetry integration with MTConnect, IO-Link, MySQL, and Python, alongside coolant-conductivity forecasting. Includes SQL ingestion, residual diagnostics, autocorrelation analysis, and prediction-interval validation. |
- ML engineering & forecasting: Python, Random Forest regression, Temporal Fusion Transformers, PyTorch Forecasting, Prophet, and SARIMAX; feature engineering, training workflows, and temporal model validation.
- Data science & statistics: time-series analysis, regression, residual diagnostics, autocorrelation, prediction intervals, MAE/RMSE/WAPE, and error analysis across operating conditions.
- Data systems & software: Python and SQL data pipelines, sensor/event-data integration, cloud-warehouse analytics, MTConnect and IO-Link interfaces, and modular forecasting and evaluation scripts.
- Applied AI & evaluation: LLM classification prompts and an ongoing study of LLM-assisted anomaly verification, threshold calibration, precision–recall tradeoffs, and scenario-based testing.
Company-related projects are documented through curated public summaries. Private data and internal artifacts are excluded; the rotating-equipment example uses synthetic values.