Now

Machine learning for drug design

• State-space (Mamba) sequence models for learning RNA representations
• Differentiable RNA folding providing secondary-structure features
• Guided discrete diffusion for generative RNA sequence design with property objectives
• Hierarchical Bayesian / beta-binomial GLMMs for sparse biological count data

Past

Off-ramp exchange for BitClout (Bitswap, co-founder)

• Designed and built a full order-book exchange moving funds between BitClout and Ethereum in seconds
• Processed millions in transaction volume

Deep probabilistic models for drug discovery (Synbiolic)

• Variational autoencoder for de novo small-molecule generation, with latent-space optimization toward QED
• Retrosynthesis pathways generated with IBM RXN
• Microsoft Imagine Cup finalist

Machine learning research (Holy Grail, YC19)

• Bayesian active learning (MC-dropout, BALD acquisition) and feature-importance estimation

Super-resolution with generative networks (NeuraScale)

• SRGAN for single-image super-resolution, with perceptual and adversarial losses
• #PoweredByTensorFlow 2.0 challenge winner

Real estate technology (AscendBuild, Next 36)

• Geospatial search engine with Mapbox GL visualizations

Coursework

Writing

• Super-Resolution
• Protein Folding
• Capsule Networks

                               ____________
                            __/ ///////// /|
                           /              ¯/|
                          /_______________/ |
    ________________      |  __________  |  |
   /               /|     | |  ty for  | |  |
  /               / |     | | visiting | |  |
 /_______________/  |/\   | |          | |  |
(_______________(   |  \  | |__________| | /   
(_______________(   |   \ |______________|/ ___/\
(_______________(  /     |____>______<_____/     \
(_______________( /     / = ==== ==== ==== /|    _|_
(               (/     / ========= === ===/ /   ////
 ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯      / ========= === ===/ /   /  /
                     <__________________<_/    ¯¯¯