Junnan Li
Senior Director at Salesforce
Singapore, Singapore
Summary
🤩
Rockstar🎓
Top SchoolJunnan Li is a senior technology leader and multimodal AI specialist with a PhD in Artificial Intelligence from the National University of Singapore and over a decade of experience spanning research, productization, and startup leadership. Currently Senior Director at Salesforce, he progressed through research and management roles there and briefly co-founded Rhymes.AI as Chief Multimodal Scientist, demonstrating both academic depth and entrepreneurial drive. His hands-on contributions to high-profile open-source vision-language work (notably code updates to Salesforce’s ALBEF repository) underscore expertise in model architecture, training pipelines, and multimodal pretraining. Based in Singapore, he blends deep research credentials with practical engineering—moving cutting-edge multimodal models from prototype to production. Colleagues know him for bridging rigorous scholarship with product-focused delivery and for preferring concrete code contributions alongside strategic leadership.
10 years of coding experience
5 years of employment as a software developer
The University of Hong Kong (HKU)Doctor of Philosophy - PhD Artificial Intelligence, Doctor of Philosophy - PhD Artificial Intelligence at National University of Singapore
Github Skills (9)
pytorch10
machine-learning10
python10
contrastive-learning10
nlp9
representation-learning9
computer-vision8
transformers8
bert8
Programming languages (2)
Github contributions (5)
salesforce/ALBEF
Jul 2021 - Sep 2022
Code for ALBEF: a new vision-language pre-training method
Role in this project:
ML Engineer
ML EngineerContributions:37 commits, 2 PRs, 24 pushes in 1 year 2 months
Contributions summary:Junnan primarily focused on updating various Python scripts related to the ALBEF model. These updates included modifications to training and evaluation scripts (`Pretrain.py`, `NLVR.py`, `Pretrain_nlvr.py`, `VE.py`), model definitions (`model_pretrain.py`, `model_retrieval.py`, `model_vqa.py`, `xbert.py`), and the `vit.py` and `tokenization_bert.py` files, suggesting an involvement in model development, training, and optimization. The changes encompass modifications to the core ALBEF model architecture and training procedures, which indicates a contribution to the central functionalities of the vision-language pre-training project.
pytorchimage-textvisionvision-languagedeep-learning
salesforce/BLIP
Jan 2022 - Sep 2022
PyTorch code for BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation
Contributions:42 commits, 5 PRs, 36 pushes in 7 months
pytorchunderstandingbootstrappingvisionvision-language
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.