Edge AI Multimodal Agentic Reasoning Engineer

Analog Devices, Inc.; San Jose (CA); On-site; USD 198,000 - 272,000; Full time

Benefits offered by this job

Medical, vision and dental coverage

401k

Paid vacation

Job summary

Analog Devices, Inc. and Lorenz Labs seek a Deep Agentic Reasoning Engineer to design multimodal reasoning models for Edge AI. Collaborate across AI, NLP, audio, and time-series teams to advance agentic systems that reason about physical-world data.

You will develop and evaluate transformer-based models, apply CoT techniques, and prototype in Python with PyTorch/TensorFlow, driving advancements in multimodal reasoning within a fast-paced research setting.

Qualifications

• PhD in AI-related fields is required.
• Strong publication record in top venues (NeurIPS/ICML/ICLR/ACL/CVPR).
• Proven experience building and evaluating transformer-based multimodal models.
• Experience with CoT reasoning and evaluation of LLMs.

Responsibilities

• Develop and evaluate deep learning models for multimodal classification and reasoning tasks.
• Engage in full research lifecycle from data collection to rigorous evaluation.
• Prototype models in Python (PyTorch/TensorFlow) and implement reasoning improvements.
• Apply distillation to transfer capabilities to efficient models.
• Document experiments for internal reports and external publications.

Skills

Transformer models

Multimodal reasoning

Chain-of-Thought

Python

PyTorch

TensorFlow

Reinforcement learning

LLM evaluation

Langchain

Langgraph

Education

PhD in AI-related fields

Tools

PyTorch

TensorFlow

Langchain

Langgraph

Job description

Analog Devices, Inc. and Lorenz Labs seek a Deep Agentic Reasoning Engineer to design multimodal reasoning models for Edge AI. Collaborate across AI, NLP, audio, and time-series teams to advance agentic systems that reason about physical-world data.

You will develop and evaluate transformer-based models, apply CoT techniques, and prototype in Python with PyTorch/TensorFlow, driving advancements in multimodal reasoning within a fast-paced research setting.