Giacomo Negri

Believe in what could be. Then build it.

I am a data scientist by training and a financial data professional by practice, working at the intersection of markets, statistics, and technology at Bloomberg in London. This site brings together the ideas, projects, and experiences acquired through my professional journey. It documents what I am building and learning along the way. You can learn more about me on LinkedIn and GitHub, or contact me at negrigiacomo02@gmail.com. Thanks for visiting. I hope you enjoy exploring! :-)

Timeline

Bloomberg

London, Data Professional

ExperienceSeptember 14, 2026

EDM-CSDI for Financial Time Series (MSc Thesis)

Conditional diffusion models for generating S&P 500 returns and reproducing their statistical properties.

ProjectJuly 20, 2026

First Half-Ironman

Independent Half-Ironman

EventJuly 18, 2026

Conditional Latent Diffusion with SDEs

Conditional face generation with continuous-time SDEs, latent diffusion, and classifier-free guidance.

ProjectFebruary 24, 2026

Writings

A New Large Language Model

Diffusion language models replace next-token prediction with iterative denoising. The result is faster, more flexible generation, and a new set of trade-offs.

October 1, 2026

Psychohistory with Modern Science

Asimov imagined a mathematics capable of steering civilization. Modern social science has built fragments of it, and discovered why the whole may be impossible.

September 1, 2026

From Noise to Data

A mathematical introduction to diffusion models, from discrete denoising and score matching to stochastic differential equations.

June 1, 2026

When the World Moves, the Prediction Should Follow

Equivariance in deep learning tells us how a neural network predictions should change when the input is transformed.

March 1, 2026

                                                
                                                
                                                
                                                
                                                
                      ..                        
       ........    .....,.,                     
    ..,-.,..~~,....,---~~~..,                   
   .,::;,--....=!....~;==!=~,.                  
   .~;=!.     ,,,.~..,:=!**==-.,                
   ,:!**.-      .,:!...=!*##*!~,,               
   ,~=###,.    ..~!!..-,=*##$#!;-.              
   ..:!#$*.,   .-=!., ..-~*#$$*==..             
    .,;!$$#!...,-..    ...:*$$$#!=-             
     ,.:=*$$#;....,.,.. ,:.!*$$$#!~,,           
      .,,:!##$#=.,,.,,,.,.,,!#$$$*!;,           
        ..,:=*#$#**!;:~~:;=,-!#$$$*=~..         
           .,,:!!*#########*..*#$$$!=-..        
             .....~:::~~-,.....*#$$#!;-         
                  ..,.....=#..,;*$$#*;~,        
                         .,**.,,!#$#*=:,        
                          ..!!,,-!###!;-.       
                           .,==..=*#*!=~.       
                            ..::..!!!=;~.       
                             ..,,,;-~:~-,       
                              ..,,..,,..        
                                 ..

Spinning (2,3)-torus knot, because is spoko.