Machine Learning Engineer

The core technology relies on fusing spectral signatures with visual and multi-sensor data to classify materials and drive precision recycling. As a Spectral ML Engineer, you will own the core classification models and build the online learning system that selects the most informative shot locations on physical materials.

What You Will Do

• Spectral Preprocessing: Own baseline correction, normalization, denoising, and derivative extraction.
• Core Classification: Develop and optimize models spanning chemometrics baselines, 1D CNNs, and transformer architectures.
• Online Learning & Decision Layer: Build, deploy, and monitor sleeping and contextual multi-armed bandit policies (e.g., UCB, Thompson Sampling) to choose optimal measurement locations under dynamic arm availability, delayed/noisy rewards, and drift.
• Multimodal Sensor Fusion: Integrate 1D spectral data with visual and real-time streaming sensor inputs into cohesive, production-grade multimodal architectures.
• Evaluation & Production: Establish rigorous offline/online evaluation frameworks and regret monitoring pipelines to push algorithms directly to physical machinery in production.

• Education: PhD or Postdoc in Physics, Astrophysics, Materials Science, or a related quantitative field.
• Experience: 0-4 years post-PhD experience (new grads accepted) focused on spectroscopy, signal processing, or applied ML with spectral data.
• Technical Mastery: Strong Python and PyTorch proficiency.
• Bandits & Online Learning: Practical experience implementing bandit algorithms (UCB, Thompson sampling, sleeping/contextual bandits) and handling classification under severe class imbalance.
• Physics Depth: Strong foundational understanding of spectral physics and 1D sensor signal processing, rather than purely high-level applied ML.

Nice to Have

• Spectroscopy or chemometrics experience with LIBS, Raman, NIR, or hyperspectral datasets.
• Hands-on experience deploying contextual bandits or reinforcement learning in live production environments.
• Familiarity with streaming systems, sensor fusion, and industrial measurement hardware.