The Sr. Applied Scientist will be responsible for building advanced time series forecasting models for Amazon Devices using deep learning techniques and state-of-the-art sequence modeling approaches. They will develop scalable, accurate predictive models leveraging RNNs, LSTMs, and Transformer architectures to drive key decision making that supports critical business decisions (including pricing, promotions and supply chain) for all of Amazon consumer hardware product lines WW. Collaborating with cross-functional teams, they will integrate these models into operational systems to enhance data-driven decision-making and optimize business outcomes. Strong expertise in time series analysis, deep learning frameworks, and computational efficiency will be key to success in this role.
Key job responsibilities
• Build and implement advanced forecasting models using deep learning techniques, with expertise in time series analysis (trends, seasonality, stationarity) and sequence modeling architectures (RNNs, LSTMs, GRUs, Transformers) to drive decisions for pricing, promotions, and supply chain optimization.
• Write efficient, scalable production code in Python, utilizing NumPy, Pandas, and deep learning frameworks (TensorFlow/Keras, PyTorch) to develop and deploy models, while collaborating with data engineers to ensure smooth integration into real-time systems.
• Continuously optimize models through sophisticated feature engineering, hyperparameter tuning, and GPU acceleration techniques, while implementing appropriate preprocessing strategies for handling missing values, outliers, and data normalization.
• Evaluate and communicate model performance using industry-standard metrics (MAE, RMSE, MAPE), create compelling visualizations using Matplotlib, and provide actionable recommendations to cross-functional teams for pricing, promotions, and supply chain strategies.
