Meta AR/VR Job | Codec Avatars Large Scale Experimentation Research Engineer

Job（岗位）: Codec Avatars Large Scale Experimentation Research Engineer

Type（岗位类型）: 3D Software Engineering

Citys（岗位城市）: Pittsburgh, PA

Date（发布日期）: 2023-7-22

Summary（岗位介绍）

Meta Reality Lab’s Codec Avatar Research team is building technology to enable immersive, photorealistic social presence. Codec Avatars are real-time live-drivable representations that match the appearance of their users. As part of the Lab’s Instant Codec Avatar group, you’ll work to scale up Codec Avatar technology by modeling the diversity of human appearance and applying that model to the process of rapidly generating new avatars.

This role is focused on our Large Scale Experimentation work, which both supports our new Research Supercluster compute resource and uses that resource to run large-scale machine learning experiments that advance the state-of-the-art in Codec Avatar technology. In this role, you will join a team of software engineers, research engineers, and research scientists to deliver software systems needed to support large scale model training over thousands of GPUs. These systems ingest, store, and serve some of the largest ML training datasets in the world, and coordinate complex workflows composed from a mixture of traditional graphics and ML algorithms. You’ll also design, execute, and analyze research experiments using those workflows to advance our understanding of how appearance modeling scales over large populations.

Qualifications（岗位要求）

Experience developing and debugging distributed systems

Bachelor’s degree in Computer Science, Computer Engineering, relevant technical field, or equivalent practical experience.

Effective communication skills, including experience driving decision making

Quantitative reasoning skills to analyze trade offs of different hardware and software solutions

Masters or higher degree in Computer Science or related technical field, or equivalent experience.

Experience working with cross functional teams including hardware, software, network, legal, privacy and security

Python experience

Proven Linux/shell scripting development experience

Experience developing reliable multi-stage data pipelines

Description（岗位职责）

Develop and debug machine learning workflows on a large multi-node cluster.

Automation of data ingress into cluster

Implement compute allocation policy for the cluster

Define and implement strategy for compute environment management and deployment

Define and communicate cluster software requirements, based on research needs

Enabling adoption of the cluster by additional research cases

Definition, design and implementation of automated testing

Point of contact for hardware & software questions regarding cluster capabilities

Reporting on progress, presenting technical risks, challenges and status to executive management.

Partner with Data Collection and Asset Generation teams to specify and ingest assets required for large scale training.

Partner with Codec Avatars Universal Avatar Research team to support large scale experimentation based on python workflows.

Partner with Research SuperCluster production engineering team to support reliable operation.

Partner with Research SuperCluster storage engineering team to support development of features required for Codec Avatars datasets

Partner with security, privacy, and policy teams to ensure workflow compliance with company policy.

Additional Requirements（额外要求）

Experience with containers (Docker or similar)

8+ years of experience

Experience developing or applying computer graphics algorithms

5+ years of experience developing workflows for large scale AI training

Experience developing or applying computer vision algorithms

Experience operating in a self-directed environment with multiple stakeholders across multiple teams

Understanding of deep neural network training

Experience with securing sensitive data (encryption, access control, audit logging)

Experience with HPC (High Performance Computing)

Experience with scheduling systems such as Slurm or Kubernetes

Large scale object storage services (S3 or similar)

Experience in research or converting research to products

git

Conda

SQL databases

C++ experience