PeopleofSuperintelligence

Researcher

Senyu Tong

Also known as Chris Tong

Senyu Tong is a research engineer whose work covers large language model training, reinforcement learning and inference systems. At Carnegie Mellon, they support graduate courses in deep learning in 2020 and machine learning in 2021. In 2025, they coauthor Apple's foundation-model report, the AXLearn training system and RLAX, a TPU framework for language-model reinforcement learning. Their 2026 work includes ECHO-2 for distributed reinforcement-learning rollouts and Parallel Track Transformers for multi-GPU inference. As of October 2026, they are a Research Engineer at Meta.

Now at
Meta
Works
1
Citations
0
h-index
0
Badges

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. 2020

    Teaching deep learning

    In Spring 2020, they serve as a teaching assistant for Carnegie Mellon's 10707 Deep Learning course, a graduate class on methods including neural networks and language models.

  2. 2021

    Advanced machine learning staff

    During Fall 2021, they join the course staff for Carnegie Mellon's 10-715 Advanced Introduction to Machine Learning, which serves PhD students pursuing machine-learning research.

  3. July 2025

    Apple Intelligence models

    In July 2025, they coauthor Apple Intelligence Foundation Language Models: Tech Report 2025. It describes a 3-billion-parameter on-device model optimized for Apple silicon with KV-cache sharing and 2-bit quantization-aware training, alongside a server model built on a Parallel-Track Mixture-of-Experts transformer.

  4. July 2025

    AXLearn training system

    A July 2025 paper introduces AXLearn, a production system for scalable, high-performance training of large deep-learning models, with Tong among its coauthors. Its modular interfaces support rapid development on heterogeneous hardware.

  5. December 2025

    RLAX preprint withdrawn

    In December 2025, they coauthor RLAX, a large-scale distributed reinforcement-learning framework for large language models on TPUs. The authors withdraw the preprint after internal stakeholders decide that the topic is inappropriate to publish at that time.

  6. February 2026

    ECHO-2 distributed rollouts

    For ECHO-2, they coauthor a February 2026 paper on distributed reinforcement learning for post-training language models with remote inference workers. The system overlaps rollout generation, model dissemination and training, with bounded policy staleness as a user-controlled setting.

  7. February 2026

    Parallel Track Transformers

    Also in February 2026, they coauthor the Parallel Track Transformers paper, which restructures transformer computation to reduce dependencies across GPUs. It reports up to a 16-fold reduction in synchronization operations and integrates the design into Tensor-RT-LLM and vLLM.

  8. As of October 2026

    Research Engineer at Meta

    As of October 2026, they are a Research Engineer at Meta.

Career

  1. NowMetaResearch Engineer
  2. 2020–2021Carnegie Mellon UniversityTeaching Assistant, Course Staff

From the biography.

Most-cited works

  1. Parallel Track Transformers: Enabling Fast GPU Inference with Reduced Synchronization2026 · 0 citations

From OpenAlex, as of 21 October 2026.

  1. 16% of AI career moves in the 2020s went from a university to a companyIncludes Senyu Tong. We estimate that 16% of AI career moves in the 2020s went from a university to a company; 29% of moves starting at a university went to a company, and 76% ended the university role.
  2. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Meta. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.
  3. Companies supply the most prior-work hires at 7 of the 9 frontier AI labsIncludes Meta. Our estimates of the organizations where 550 people worked directly before joining 9 frontier AI labs across 620 career transitions, comparing the main feeder organizations per lab.
  4. Of 943 people shaping AI who worked at Google, 446 work there nowIncludes Meta. Our estimates of where the 943 people shaping AI who worked at Google and Google DeepMind work now: 446 at Google and 452 across 278 other organizations.

Colleagues

People with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Senyu Tong yet. Send us a free-licensed photo.

  1. 10707 - Deep Learning, Spring 2020 andrejristeski.github.io. Checked 21 October 2026.
  2. 10-715 Fall 2021: Advanced Introduction to Machine Learning cs.cmu.edu. Checked 21 October 2026.
  3. Apple Intelligence Foundation Language Models: Tech Report 2025 arxiv.org. Checked 21 October 2026.
  4. AXLearn: Modular Large Model Training on Heterogeneous Infrastructure machinelearning.apple.com. Checked 21 October 2026.
  5. RLAX: Large-Scale, Distributed Reinforcement Learning for Large Language Models on TPUs arxiv.org. Checked 21 October 2026.
  6. ECHO-2: A Large-Scale Distributed Rollout Framework for Cost-Efficient Reinforcement Learning arxiv.org. Checked 21 October 2026.
  7. Senyu Tong - Google Scholar scholar.google.com. Checked 21 October 2026.
  8. Parallel Track Transformers: Enabling Fast GPU Inference with Reduced Synchronization arxiv.org. Checked 21 October 2026.
  9. Parallel Track Transformers: Enabling Fast GPU Inference with Reduced Synchronization machinelearning.apple.com. Checked 21 October 2026.