I am a Senior Machine Learning Engineer at Meta Superintelligence Labs (MSL), where I architect core infrastructure for frontier-scale multimodal AI models and large language models. I hold a Master of Science in Computer Engineering from NYU Tandon School of Engineering and a Bachelor of Engineering in Computer Science from Harbin Institute of Technology.
Current interests & recent work include:
- Multimodal AI Infrastructure, Data Loading Pipelines, and Large-Scale Model Training [Llama 3 Herd of Models 2024]
- Direct Preference Optimization (DPO) for LLM Alignment and Fine-Tuning [Meta AI Research 2024]
- ML Ads Systems, Ranking Optimization, and Automated Machine Learning [Meta Ads ML 2024]
- Distributed Systems, Graph Algorithms, and Database Query Processing [ISJ 2021, ICIS 2018]
I'm open to collaboration opportunities on AI infrastructure, ML systems, and frontier-scale model development—feel free to connect!
News
- [03/25] Contributing to TorchTune, PyTorch's native library for LLM fine-tuning with LoRA/QLoRA support.
- [12/24] Acknowledged in "The Llama 3 Herd of Models", a groundbreaking AI research publication cited 4,000+ times!
- [03/24] Led Multi-modal DPO implementation, enabling alignment for text, image, and speech modalities at scale.
- [06/23] Engineered Traffic Shift Optimization (TSO) automation saving ~1,250 engineering hours and 2% GAS gain.
- [06/20] Completed Master's degree at NYU Tandon; Joined Meta as Software Engineer in Ads Infrastructure.
Selected Publications
-
The Llama 3 Herd of Models [Paper]arXiv 2024
-
GPU-based Efficient Join Algorithms on Hadoop [Paper]The Journal of Supercomputing 2021
-
Parallel Algorithms for Flexible Pattern Matching on Big Graphs [Paper]Information Sciences 2018
Service
Reviewer & Editorial Board
-
NeurIPS 2025 Workshop BERT2S — Reviewer
Reviewing papers on BERT and semantic similarities for workshop submissions. -
ACL ARR 2025 (February Cycle) — Reviewer
Reviewed and managed multiple submissions across domains: Generation, Information Extraction, Information Retrieval, Text Mining, and Language Modeling. Active contributor across all 2024 cycles. -
ICLR 2025 Workshop AI4CHL — Reviewer
Reviewed papers on AI for Clinical Health, including novel neural ODE approaches for medical imaging and RL frameworks for healthcare applications. -
ICLR 2025 Workshop FM-Wild — Reviewer
Reviewed papers on foundation models in the wild, including privacy-preserving LLM fine-tuning systems. -
ICLR 2025 Workshop DeLTa — Reviewer
Reviewed papers on distillation and knowledge transfer, including one-step distillation frameworks for generative models and mixture-of-experts approaches.
Recognition & Acknowledgments
-
ACL ARR 2025 February — Letter of Recognition
Requested and recognized for service as a reviewer and area editor across multiple cycles in 2024.