Software Engineer III

Meta Platforms • Redmond, OR

1d • Onsite

About The Position

We are seeking a contract ML Software Engineer to bring state-of-the-art machine learning from research into real-time operation on wearable embedded devices. The role works at the intersection of model training and embedded deployment: training and adapting models alongside research scientists, then optimizing, porting, integrating, and profiling them on resource-constrained hardware. The position requires balancing model architecture, quantization, and target hardware constraints to meet latency, memory, power, and compute budgets. Familiarity with audio ML, computer vision ML, or audio DSP is a strong plus because many models use multimodal audio and visual sensor data. The team values ownership, communication, documentation, collaboration, self-motivation, continuous learning, and comfort with the ambiguity of frontier AR/VR work.

Requirements

• Bachelor’s degree in Computer Science, Computer Engineering, Electrical Engineering, or a related field, or equivalent work experience.
• Industry experience optimizing and running ML inference on embedded or resource-constrained devices.
• Experience training ML models using deep learning frameworks such as PyTorch or TensorFlow.
• At least 2 years of industry experience writing C/C++.
• Experience profiling and debugging performance issues across an ML model and its target hardware.
• Experience working in cross-functional teams.

Nice To Haves

• Experience with audio ML, such as speech enhancement, source separation, keyword spotting, or acoustic event detection.
• Experience with computer vision ML, such as detection, tracking, segmentation, or pose estimation.
• Experience with audio DSP fundamentals, such as filtering, STFT/filterbanks, beamforming, or real-time audio processing.
• Experience with on-device inference frameworks and toolchains such as ExecuTorch, TFLite/LiteRT, ONNX Runtime, or vendor NPU/DSP SDKs.
• Experience with model compression techniques such as quantization-aware training, post-training quantization, pruning, or distillation.
• Experience writing hardware-specific optimizations such as SIMD/NEON, DSP intrinsics, or fixed-point arithmetic.
• Experience in the AR/VR or wearable device space.
• Ability to gather requirements and work closely with researchers to develop novel solutions.
• History of writing code to support the execution of research initiatives.

Responsibilities

• Work with research scientists to move ML models from research prototypes to efficient, real-time inference on embedded devices.
• Train, fine-tune, and adapt ML models with latency, memory, power, and compute constraints in mind.
• Optimize models for on-device inference using techniques such as quantization, pruning, distillation, operator fusion, and architecture adjustments.
• Port, integrate, and profile ML models on embedded targets, including DSPs, NPUs, microcontrollers, and mobile SoCs, and identify and resolve performance bottlenecks.
• Write clean, maintainable, high-performance C/C++ code for on-device inference pipelines and integrate it into real-time prototype systems.
• Build tools and infrastructure to help researchers evaluate model performance and accuracy on target hardware.
• Coordinate with internal hardware, runtime, and toolchain teams across Reality Labs to evaluate and integrate with existing systems.
• Collaborate as part of a multifunctional team, contributing communication, documentation, and teamwork.
• Learn new areas and technologies and work through the ambiguity of AR/VR problem solving.

What This Job Offers

• Job Type: Full-time
• Career Level: Mid Level