About

Amir Ziai is a Senior Machine Learning Engineer at Netflix, where he is responsible for building scalable ML systems for video understanding. With over 2.69 years of relevant experience, Amir has a diverse background in ML, deep learning, NLP, data science, and research science. At Netflix, Amir is currently building multimodal ML systems for temporal video tagging at scale. He has designed an event-driven system for scoring videos using an extensible set of models and built a human-in-the-loop framework for evaluating the quality of video models. Amir has also trained multimodal video representation models and built a library for rapidly training and deploying zero-config video moment classifiers. Prior to his current role, Amir worked as a Senior Data Scientist at Netflix, where he built a system for enabling security experts to rapidly hypothesize and deploy heuristics for detecting anomalous activity. He also developed a time series anomaly detection tool using a language modeling framing and significantly improved the scalability of an existing model that allowed for 2x the number of features with a quarter of the original compute footprint. Amir created tooling for detecting train/test skew and metric adjustment and developed a library for streamlining experimentation with new features and model architectures/hyperparameters. Amir has also worked as an XCS224N Course Assistant at Stanford University, where he taught NLP and deep learning. He holds a Graduate Courses degree in ML and systems from Stanford University, an MS in Computer Science and ML from Georgia Institute of Technology, an MS in Data Science from the University of California, Berkeley, an MS in Intelligent Systems & Control Engineering from Simon Fraser University, and a BS in Mechanical Engineering from Sharif University of Technology. Amir is proficient in Scala, Python, Spark, TensorFlow, PyTorch, scikit-learn, and XGBoost. He is passionate about using his expertise to build innovative ML systems that can make a positive impact on people's lives.