Research Scientist Intern - Neural Representations Compression

Meta • Redmond, WA

1d • $7,650 - $12,667

About The Position

The intern will research compression and streaming techniques for neural representations used in split-inference AI and AR/VR systems. The work focuses on compact, efficient exchange of internal neural model states, including latent tensors and 3D Gaussian splats, where compression affects latency, bandwidth efficiency, battery performance, memory footprint, and fidelity.

The intern will explore latent vector compression through quantization, entropy coding, and learned prior models, as well as compression and streaming algorithms for Gaussian splats that preserve spatial distribution, opacity, and color parameters. The project includes reviewing existing research, implementing prototype models, and benchmarking them against traditional compression methods. The goal is to develop a neural compression framework that improves compression ratio, speed, and fidelity, with applications in AR/VR systems and other split-inference applications.

Requirements

• Currently has or is in the process of obtaining a PhD in Image and Video Processing, Signal Processing, Information Theory, Machine Learning, Computer Graphics, or a related field.
• Coding experience in C/C++ and/or Python.
• Experience in image and video processing.
• A proven track record of significant results, demonstrated through grants, fellowships, patents, or first-authored publications at leading workshops or conferences such as ICIP, DCC, CVPR, IEEE Transactions or Letters, or ACM SIGGRAPH.

Nice To Haves

• Intent to return to a degree program after completing the internship/co-op.
• Knowledge of applications related to neural networks and compression.
• Knowledge of existing image and video codec standards, such as H.26x, VPx, and JPEG.
• Knowledge of existing Point Cloud compression standards, such as MPEG PCC.
• Knowledge of learning-based compression.

Responsibilities

• Explore issues involved in processing, compressing, and streaming neural and/or learned representations.
• Identify, with assistance from mentor(s), opportunities to significantly improve compression efficiency and/or quality.
• Work with peers to build a working prototype of the proposed algorithm.
• Communicate ongoing research progress and final results.
• Review existing literature, implement prototype models, and benchmark them against traditional compression methods.

What This Job Offers

• Job Type: Full-time
• Career Level: Intern
• Education Level: Ph.D. or professional degree