Satar Shamsi

I've built and deployed an end-to-end LSTM forecasting system for AAPL stock prices, covering preprocessing, training, validation, experiment tracking, and live inference. The model achieved 3.11% RMSE on a held-out test set.

I integrated MLflow for experiment tracking, model versioning, and Model Registry workflows; used Optuna for hyperparameter optimization; and fixed a preprocessing data-leakage issue that had inflated validation performance. I also modularized the system into reusable OOP components, containerized training and inference with Docker, and automated testing and CI/CD through GitHub Actions.

At GoalEarn, I'm completing a part-time ML/AI engineering internship across LSTM model evaluation, ETL data engineering, and inference API development. I typically complete assigned project tasks independently and ahead of their allotted timelines.

My graduate research at ITMO University included machine learning-based music generation and SEIR mathematical modeling of epidemic dynamics, supported by a foundation in electrical engineering.