We are seeking a creative and hands-on Deep Learning Research Engineer to drive model and architecture innovation across our sensing platforms. In this role, you will go beyond off-the-shelf models, exploring novel neural network designs, designing rigorous training and validation pipelines, and translating inventive concepts into clean, high-performance implementations. While our domain deals with physical data, this is an ML-first role: we are looking for a deep learning specialist with strong research intuition, expertise in PyTorch, and the drive to solve non-standard, real-world problems.
Responsibilities
Conceive, design, and implement novel neural network architectures tailored for complex physical sensing, reconstruction, and inverse problems.
Lead end-to-end hands-on research: data preparation, model formulation, training, hyperparameter exploration, and failure-mode analysis.
Build, train, and rigorously validate deep learning models using Python and PyTorch.
Formulate custom loss functions, regularization techniques, and inductive biases that capture domain-specific constraints.
Design comprehensive validation and benchmarking suites to stress-test model robustness, generalization, and scaling behavior.
Propose creative, non-conventional modeling ideas, stay abreast of state-of-the-art DL literature, and rapidly prototype proof-of-concepts.
Collaborate cross-functionally with physicists and optical software engineers to benchmark models against simulated and experimental lab datasets.
Requirements: Minimum Qualifications
M.Sc. in Computer Science, Electrical Engineering, Data Science, Applied Mathematics, Physics, or a related technical field.
3+ years of hands-on research experience in machine learning and deep learning (in industry R&D or an academic research lab).
Deep theoretical and practical understanding of modern deep learning fundamentals (e.g., neural network architectures, optimization landscapes, generalization, attention/transformers, CNNs, or implicit representations).
Proven hands-on proficiency in Python and PyTorch, including writing custom layers, loss functions, and efficient training loops.
Demonstrated track record of creative problem-solving, proposing novel ideas, and executing them from hypothesis to experimental validation.
Strong technical communication skills and fluency in English
Preferred Qualifications
Track record of research contributions (e.g., publications in top-tier conferences like CVPR, ICCV, NeurIPS, ICLR, ICML, or relevant patents).
Foundational understanding or interest in physics, wave optics, signal processing, or computational imaging.
Experience with physics-informed neural networks (PINNs), neural operators (FNO/DeepONet), or integrating physical differential equations into loss functions.
Experience with large-scale distributed training, model profiling, and GPU acceleration.
This position is open to all candidates.