Applied Researcher: On-Device Multimodal Reasoning

What you'll do

• Design and train compact vision-language models for multi-step visual reasoning within on-device resource budgets.
• Develop efficient reasoning, decoding, compression, and perception-expert integration methods.
• Optimize models for Apple silicon and evaluate reasoning quality against latency, memory, and power budgets.

What they ask for

• MS in Computer Science, Machine Learning, AI, Computer Vision, or a related field (or equivalent practical experience).
• Strong foundation in deep learning, with specific experience in LLM or VLM training, post-training, or inference optimization.
• Demonstrated experience with multimodal models in resource-constrained environments, including hands-on work with reasoning quality, decoding, or model compression.
• Proficiency in Python and modern deep learning frameworks; familiarity with inference and optimization toolchains.

Nice to have

• PhD with research in efficient multimodal reasoning, LLM reasoning, model compression, efficient inference/decoding, or lightweight VLM architectures.
• Experience training or post-training vision-language models end-to-end.

Pay

USD $150,400 to $277,600 base salary

Tools

Python, PyTorch, vLLM/SGLang, llama.cpp, MLX, CoreML, GRPO, RLVR

AI focus

Efficient on-device multimodal reasoning: compact vision-language models that reason over images, video, 3D scenes, and outputs from specialist perception models, with compressed reasoning and device-aware inference.