   
   
  
shawn@renormtech.com


Strong leadership             
Comprehensive engineering knowledge         
            
          
             
 
Solid theory background              
            
                  
    
Fast learner                
              

      
  
           
        
      
  
       
          
 
     
            
 
   
      
  
  
   
 
 
       
  
 
 

  
 

   
     
    
   
Projects

               
     
PERSIA supports training models with over 100 trillion parameters in production environments.  
               
         
            
      

 more than a million videos uploaded per hour           
               
               
     
                
    
      
      

                    
                
              
             
   
               
     
    

                  
  
            
              
       
     
    
    
Projects
        
            
              
                  
                 
     

             
          
     
    
Projects
     
                 
             
                 
       
               
               
           
 

     
      

  
  

  

  
         
         
         

 

 
         
        
      

  
         
        

    

   
    
               
               
    Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery & Data
Mining 
             
                 
      VLDB 
               
              
               
               
             3rd IEEE
International Conference on Artificial Intelligence Circuits and Systems, AICAS 2021
                
       International Conference on Machine Learning

              
             
   International Conference on Machine Learning 
               
         Advances in Neural Information
Processing Systems 
         DoubleSqueeze    
     International Conference on Machine Learning 
             
   Proceedings of Machine Learning Research      
    
           D
2
   
  International Conference on Machine Learning 
             
  International Conference on Machine Learning   
               
          
 Advances in Neural Information Processing Systems     
             
  International Conference on Artificial Intelligence and Statistics 
               
      Advances in Neural Information Processing Systems 
              
           Advances in
Neural Information Processing Systems 
              
  International Joint Conference on Artificial Intelligence 
              
  Advances in Neural Information Processing Systems     
 
                
      (
0.8

0.2
)   Physical Review
B   

       Stochastic Recursive Variance Reduction for Efficient Smooth Non-
Convex Compositional Optimization  
1912.13515 [stat.ML]
               APM-
Squeeze: A Communication Efficient Adam-Preconditioned Momentum SGD Algorithm   2008 .
11343 [cs.DC]
         Stochastic Recursive Momentum for Policy Gradient
Methods   2003.04302 [stat.ML]
 
 
 
     

        
       
        
        
       
   
           
    
       
    
       
      
      
    
    
 
       
    


 
 
 
   
   
     
    
  
     
 

      
           
 

    Bagua! Distributed Communication Library

  Technology best practices - Large scale GPU based learning system for ad recommendation
  DoubleSqueeze: parallel stochastic gradient descent with double-pass error-compensated com-
pression
  On model compression and distributed training of the Face Landmarks model

    Asynchronous Parallel Empirical Variance Guided Algorithms for
the Thresholding Bandit Problem
  Asynchronous Decentralized Parallel Stochastic Gradient Descent
  D2: Decentralized Training over Decentralized Data

        Accelerating Deep Learning via Decentralized Par-
allel Optimization

    Asynchronous Parallel Stochastic Gradient for Nonconvex Optimization
 
   
   
       

                
                
  
               
             
 
 
       
               
                  
   
         
 
           
            
   
      
 

         
      
         
    
 
      
     
       
           io_uring 
epoll          
       no_std
         
 
       
          
  
  
 
     
     
       