Carlos Xavier Hernández
Senior Research Scientist · London, UK
🐙 github |
💼 linkedin |
🌐 website |
✉️ email
I am a Senior Research Scientist at Meta Reality Labs, now based in London, with 8+ years of experience building machine learning for consumer wearables. My work on wearable neuromotor interfaces spans the full pipeline, from biosignal processing and multimodal sensor fusion (sEMG and IMU) to training, evaluating, and deploying deep learning models in production. Prior to that, I worked with Vijay Pande at Stanford on probabilistic models of biomolecular dynamics.
Experience
Senior Research Scientist · New York, NY, USA · 2019 – Present
- Shipped gesture recognition to consumers as the technical lead of a team of 8+ research scientists and engineers developing for the Meta Neural Band (launched Sep 2025), by designing and training deep learning models that decode real-time multimodal signals into discrete input controls on-device for Meta Ray-Ban Display.
- Achieved >90% gesture classification accuracy on held-out users without the need for individual calibration, by architecting a generic LSTM-based neural decoding model trained on large-scale sEMG datasets collected from ~5,000 participants. Co-authored peer-reviewed publication in Nature demonstrating the first high-bandwidth non-invasive neuromotor interface with cross-user generalization (0.88 gestures/sec in closed-loop tests with first-time users), contributing core ML model development and evaluation methodology.
- Demonstrated viability of EMG-based controls for users with hand tremor (featured at Meta Connect 2024), by leading cross-functional accessibility data collection and analysis to show that EMG-based models can accurately decode motor intent despite involuntary movement artifacts, achieving >80% gesture classification accuracy on the population with hand tremor.
Research Scientist · New York, NY, USA · 2018 – 2019
- Built production-grade ML training and inference pipelines for personalization of real-time gesture recognition models, by developing end-to-end data processing, training, and fine-tuning infrastructure for wrist-based sEMG decoding.
- Established foundational R&D for EMG-based neural interfaces prior to acquisition, by conducting early research on time-series signal processing and deep learning approaches for decoding motor signals into user intent.
Stanford University
NSF Graduate Research Fellow · Stanford, CA, USA · 2013 – 2018
- Authored peer-reviewed publication in Phys. Rev. E describing the Variational Dynamical Encoder (VDE), a time-lagged variational autoencoder that compresses high-dimensional time-series into a single interpretable, low-dimensional latent representation. The VDE retained over twice the mutual information with input features compared to linear methods, and on protein folding simulations resolved a slowest dynamical process 2× longer than the leading linear baseline (tICA).
- Co-developed open-source scientific computing tools widely adopted across the computational biology community, by building Python libraries for molecular dynamics trajectory analysis, Markov state modeling of biomolecular kinetics, and automated hyperparameter optimization.
- Achieved an R² of 0.987 and MSE of <0.1 in automated cell counting and segmentation, by training a convolutional neural network pipeline (FPN + VGG-11) with uncertainty estimation on ~10,000 microscopy images, replacing a time-intensive manual process with computer vision.
Education
Stanford University
Ph.D. in Biophysics · Stanford, CA, USA · 2013 – 2018
Advisor: Vijay Pande
Columbia University in the City of New York
B.S. in Applied Mathematics · New York, NY, USA · 2009 – 2013
Skills
Languages & Frameworks: Python, PyTorch, NumPy, SciPy, Pandas
Domains: Time-series modeling, biosignals, signal processing (DSP), causal inference
Methods: Deep learning (RNNs, Transformers), large-scale distributed training, fine-tuning, Markov state models, information theory
Selected Publications
A Generic Non-Invasive Neuromotor Interface for Human-Computer Interaction
P Kaifosh, TR Reardon, and CTRL-labs · Nature · 2025
📚 136
Variational Encoding of Complex Dynamics
CX Hernández*, HK Wayment-Steele*, MM Sultan*, BE Husic, and VS Pande · Phys. Rev. E · 2018
📚 152
Using Deep Learning for Segmentation and Counting within Microscopy Data
CX Hernández, MM Sultan, and VS Pande · arXiv · 2018
📚 37
Selected Software
MDTraj: A Modern, Open Library for the Analysis of Molecular Dynamics Trajectories
RT McGibbon, KA Beauchamp, MP Harrigan, C Klein, JM Swails, CX Hernández, CR Schwantes, LP Wang, TJ Lane, and VS Pande · mdtraj/mdtraj
Python · ⭐ 736 · 🍴 298
VDE: Variational Dynamical Encoder for Complex Dynamics
CX Hernández, HK Wayment-Steele, MM Sultan, BE Husic, and VS Pande · msmbuilder/vde
Python · ⭐ 189 · 🍴 42
MSMBuilder: Statistical Models for Biomolecular Dynamics
MP Harrigan, MM Sultan, CX Hernández, BE Husic, P Eastman, CR Schwantes, KA Beauchamp, RT McGibbon, and VS Pande · msmbuilder/msmbuilder
Python · ⭐ 168 · 🍴 94
MolEncoder: Molecular Autoencoder in PyTorch
CX Hernández · cxhernandez/molencoder
Python · ⭐ 94 · 🍴 18
Osprey: Hyperparameter Optimization for Machine Learning
RT McGibbon, CX Hernández, MP Harrigan, S Kearnes, MM Sultan, S Jastrzebski, BE Husic, and VS Pande · msmbuilder/osprey
Python · ⭐ 72 · 🍴 23
Posters & Presentations
Neural Control of Movement Society Meeting
Poster · Panama City, PAN · 2025
“Stable Control through sEMG Input: Hand Gesture Recognition on a Population with Hand Tremor”
Convolutional Neural Networks for Visual Recognition (CS231N)
Invited Presentation · Stanford, CA, USA · 2017
“Using Deep Learning for Segmentation and Counting within Microscopy Data”
Biophysical Society Meeting
Poster · Los Angeles, CA, USA · 2016
“Intrinsic Disorder in the P53 C-Terminal Regulatory Domain Yields Multiple Pathways for Folding-Upon-Binding”
Workshop on Molecular and Chemical Kinetics
Poster · Berlin, DEU · 2015
“Inferring Causality Along Transition State Pathways”
Honors & Awards
- Graduate Research Fellowship · National Science Foundation · 2013
- ADVANCE Summer Research Fellowship · Stanford University · 2013
- EXROP Undergraduate Research Fellowship · Howard Hughes Medical Institute · 2012
- Genentech Summer Undergraduate Research Fellowship · Columbia University · 2011
Press
