Dinar Orazgaliyev

• Architected Forcateri, a modular PyTorch-based MLOps framework that standardized the end-to-end training and evaluation lifecycle for time-series models. • Engineered a high-performance ensemble (TFT, TCN, XGBoost, CatBoost) achieving a 11.3% MAPE, delivering a 10% accuracy lift over established baselines. • Integrated ClearML for robust experiment tracking and automated hyperparameter optimization (HPO), significantly reducing manual iteration cycles.

• Developed SQL-based validation and monitoring for medical data pipelines handling 10k+ daily records. • Identified data inconsistencies and anomalies using BigQuery, improving pipeline reliability and debugging speed.

Responsible for maintenance activities, which consist of programming DCS systems, troubleshooting problems, and working closely with Process engineers while developing improved processes. Also, responsible for the stable integration of software with other IT applications, networks, and databases. Writing SQL queries to fetch alarms.

Responsibilities include the following deliverables: servers and network switch configurations, active directory administration and server virtualization. Was responsible for applications and HMI configurations during Foxboro DCS upgrade projects.

Was responsible for historical data migration from Oracle database to MSSQL database, and developing automated web-reports on SSRS and day-to-day loaders. Developed the queries for OLAP and OLTP.