Principal AI/ML Engineer, Time-Series & Sensor Reasoning Models

About the role

Analog Devices is seeking a Principal AI/ML Engineer to build time-series and sensor foundation models for its Lorenz Labs team. The role develops architectures that unify multimodal sensor data (electrical, audio, motion, photonic and physiological signals) into a foundation for context-aware reasoning across time, contributing to the Faraday suite of physically-intelligent reasoning models. It works on multi-modal time series reasoning models using state-of-the-art techniques in time series embeddings, cross-attention, reinforcement learning and time series agentic solutions.

Key responsibilities

• Lead R&D on intelligent time-series agents for the edge, combining anomaly detection, reasoning and forecasting foundation models that handle electrical, audio, motion, physiological and text data.
• Advance sensor fusion research and create benchmarking pipelines for cross-domain time-series models covering robustness, interpretability and hardware performance.
• Apply LoRA, Q-LoRA, adapter tuning and contrastive alignment, and investigate DPO and RLAIF for physical and sensory reasoning.
• Partner with hardware, signal processing and systems teams to co-design real-time, energy-efficient sensing architectures.
• Design statistical experiments to collect sensor data and publish ADI at venues such as NeurIPS, ICLR, ICML, ICASSP and KDD.
• Mentor junior researchers and help shape the strategy for foundation models that reason about physical systems.

Required profile

• 15+ years of experience developing software products with a strong AI/ML focus, and 6+ years developing AI/ML products.
• Ph.D. in Electrical Engineering, Computer Science or Applied Physics preferred.
• Demonstrated leadership and agility in combining technical solutions to business problems, preferably for embedded systems.
• Record of innovation through patents, publications or open-source contributions.
• Willingness to travel about 10% of the time; first shift, days.

Required skills

• Deep expertise in time-series ML, signal processing and foundation models (Chronos, TimesFM, TimeGPT), including training, fine-tuning and evaluation.
• Representation learning, time series encoding, compression and motif discovery.
• Parameter-efficient fine-tuning (LoRA/Q-LoRA) and reward-based optimization (DPO, PPO, RLAIF).
• Statistical hypothesis testing, experimental design and causal discovery.
• Fluent Python, PyTorch and large-scale training pipelines on cloud or distributed systems (AWS, GCP).

What we offer

• Expected wage range of $260,360 to $357,995 for a new hire, with possible variation.
• Discretionary performance-based bonus.
• Medical, vision and dental coverage, 401k, paid vacation, holidays and sick time.