Data Scientist Direct Hire-2749

• Process and model univariate and multivariate time series; perform augmentation, segmentation, alignment, and feature engineering as needed
• Work with diverse sensor modalities (e.g., vibration, temperature, accelerometers, audio, images) including synchronization, sampling-rate challenges, and noise/artifact handling
• Build robust pipelines for ingesting and cleaning large-scale, time-synchronized multi-sensor datasets

Foundation Models & Deep Learning

• Develop self-supervised / semi-supervised methods (masked modeling, contrastive learning, temporal predictive coding, multimodal alignment/fusion)
• Design and evaluate architectures such as Transformers, CNNs (1D/2D/3D), TCNs, RNN/LSTM/GRU, and optional GNNs / diffusion / generative models
• Fine-tune at scale using domain adaptation, few-shot learning, and adapter/prompt strategies where applicable
• Define evaluation strategies across ML metrics (MSE, F1, AUC) and time-series/event metrics (DTW, correlation, IoU) tied to end-user outcomes

Software & Infrastructure

• Build training and inference workflows for multi-GPU / multi-node environments (mixed precision, scalable data loaders, distributed training/optimization)
• Contribute to high-performance data preprocessing (Python; C++/CUDA a strong plus)
• Implement production-ready data/model packaging for downstream teams and deployment

Collaboration & Communication

• Work cross-functionally with domain experts, product stakeholders, and engineering teams
• Present model behavior clearly (interpretability/attention analysis, uncertainty quantification) and communicate value/impact

Work Authorization: Must be authorized to work in the U.S. (no sponsorship) Experience Level: 3+ years (MS/PhD preferred), * 3+ years of relevant experience in Data Science / Machine Learning / AI

• Strong experience with time series / sequential data and real-world sensor challenges
• Expertise in Python (NumPy, SciPy, Pandas)
• Deep learning experience with PyTorch (preferred) and/or TensorFlow/Keras, JAX/Flax
• Understanding of core math foundations (linear algebra, probability, optimization) and signal processing basics
• Ability to communicate effectively with technical and non-technical stakeholders
• Authorized to work in the U.S. and willing to relocate

Preferred / Strongly Valued

• MS or PhD in CS, Data Science, AI, or related field
• Experience building self-supervised foundation models for time series
• Multi-modal learning (time series + vision/text/audio/structured) and sensor fusion experience
• Distributed training at scale (ZeRO, mixed precision, multi-node)
• C++/CUDA and high-performance preprocessing
• Signal processing methods (Fourier/wavelets, Kalman/Savitzky-Golay, resampling, noise modeling) #LI-DNI