Edge ML Engineer: NPU Model Optimization

ByteDance; San Jose (CA); On-site; USD 212,800 - 450,000; Full time

Benefits offered by this job

Medical insurance

Dental insurance

Vision insurance

401(k) with match

Parental leave

Disability coverage

Life insurance

Wellbeing benefits

Paid holidays

Paid sick days

Paid personal time

Job summary

ByteDance seeks an Edge ML Software Engineer for the Pico project in San Jose. The role focuses on converting and compiling ML models for edge NPUs, with emphasis on hardware-aware optimizations and quantization techniques.

You will profile power and performance across simulators and silicon, identifying bottlenecks in compute and memory workflows. The candidate should have strong Python/C++ skills and hands‑on experience with PyTorch and TensorFlow, plus a background in CNNs/Transformers.

Qualifications

• Master's degree or equivalent practical experience in a related field.
• 3+ years of industry experience in ML software engineering, model deployment, or ML systems for production environments.
• Strong understanding of deep learning architectures including CNNs and Transformers.
• Knowledge of ML accelerators' architectures, operator fusion, memory hierarchies, and data movements.
• Practical experience with PyTorch and TensorFlow.
• Proficiency in Python and C/C++.

Responsibilities

• Convert and compile ML models for edge NPUs, applying quantization mechanisms.
• Profile and analyze model performance and power consumption on simulators, emulators, and silicon.
• Identify bottlenecks related to compute, memory bandwidth, data movement, and scheduling.
• Apply hardware-aware optimization strategies, such as quantization, compression and operator fusion, to meet latency, memory and power targets.
• Collaborate with algorithm, compiler, firmware and hardware teams to debug functional and performance issues.

Skills

CNNs/Transformers

Python

C/C++

PyTorch

TensorFlow

Education

Master's degree in CS/EE/CE or related field

Tools

PyTorch

TensorFlow

Job description

ByteDance seeks an Edge ML Software Engineer for the Pico project in San Jose. The role focuses on converting and compiling ML models for edge NPUs, with emphasis on hardware-aware optimizations and quantization techniques.

You will profile power and performance across simulators and silicon, identifying bottlenecks in compute and memory workflows. The candidate should have strong Python/C++ skills and hands‑on experience with PyTorch and TensorFlow, plus a background in CNNs/Transformers.