Machine Learning Engineer - Representation Learning

About the job

We’re looking for a Machine Learning Engineer in NYC with deep expertise in self-supervised learning and large-scale representation learning. This role is for someone who wants to own the core modeling layer of a technology platform built from the ground up, turning large amounts of raw, unlabeled sensing data into general-purpose models. You’ll work directly with the founding team across machine learning, sensing, data, software, and product. This is a full-time, on-site role in New York City.

The Focus

Build AI models that learn general-purpose representations from large-scale data.

What You Will Do

• Design, train, and scale self-supervised pretraining approaches (masked modeling, contrastive, predictive/JEPA-style, and similar) for multimodal time-series data.
• Build models that generalize across users, devices, and real-world conditions.
• Develop fine-tuning and adaptation strategies for various downstream tasks.
• Define evaluation protocols and benchmarks, including linear probing, few-shot transfer, and held-out-user generalization.
• Work with the data team to shape data collection, curation, and pretraining corpus strategy.
• Partner with hardware and sensing teams so that modeling insights feed back into sensor design.
• Help build training infrastructure, experiment tracking, and engineering practices as the team grows.

What We Are Looking For

• MS or PhD in Computer Science, Electrical Engineering, Machine Learning, or a related field, or equivalent industry experience.
• 3+ years of hands-on deep learning experience, with direct work on self-supervised learning, representation learning, or foundation model pretraining.
• Strong proficiency in PyTorch and in modern architectures for sequential data (Transformers, state-space models, temporal CNNs).
• Experience training models at scale: distributed training, mixed precision, data pipelines for large datasets, and debugging training instability.
• Comfort working with real-world time-series data, noisy signals, and imperfect or sparse labels.
• Ability to move quickly from research ideas to working systems while keeping engineering rigor.
• An ownership mindset, practical problem-solving skills, and comfort with the pace and uncertainty of an early-stage startup.

Nice to Have

• Experience working with wearable sensing modalities.
• Experience with multimodal learning or cross-modal alignment.
• Publications in venues such as NeurIPS, ICML, ICLR, CVPR, IMWUT/UbiComp, or related venues.

The Perks

Direct collaboration with the founding team, a competitive salary and equity, comprehensive benefits, top-tier VC backing, and a rare chance to help build a “Zero to One” technology company at the intersection of AI and real-world human behavior understanding.

Compensation: 160K - 200K base salary + equity

Location: In-person, New York City