Hi, I’m Walid.

I work on problems in machine learning and applied mathematics.

I’m currently a machine learning (research) engineer at Reverie Labs, an early-stage hybrid of tech and biotech. I work on building our machine learning platform for computational drug discovery. On the modeling side, this spans everything from message-passing (graph) neural networks, language models, generative models and voxel-based methods, to more traditional computational chemistry tools such as docking and molecular dynamics simulations. Other methods I enjoy exploring include meta-learning and reinforcement learning, and I’ve previously worked on computer vision applications as well. I am also, broadly speaking, interested in tech / startups / finance and their intersection.


Education 📚

University of Oxford

MSc. Mathematical Modelling and Scientific Computing

Columbia University

BSc. Applied Mathematics

Publications and Preprints 🖊️

Tie-Decay Temporal Networks in Continuous Time and Eigenvector-Based Centralities
Walid Ahmad, Mason A. Porter, Mariano Beguerisse-Díaz

IEEE Transactions on Network Science and Engineering (2021)

Evaluating chemical descriptors with the loss-data framework
Walid Ahmad

Machine Learning for Molecules Workshop @ NeurIPS 2020


Posts 📃

Filtering Noisy Data

Reverie Labs blog post about using ensemble filters to detect noisy data.


Projects ⚒️

ChemBERTa

Ongoing project in collaboration with other Reverie and DeepChem team members, investigating large-scale self-supervised pretraining methods for attention-based language models. Models shared via HuggingFace 🤗. Builds on work previously shared at the ML4Molecules workshop at NeurIPS 2020.

Virtual Kinome Profiling with Message Passing Neural Networks

Reverie Labs poster presented at the Computer Aided Drug Design (CADD) Gordon Research Conference 2019.

The Database of Modern Exhibitions (DoME): European Paintings and Drawings 1905–1915

I contributed to the DoME project at the University of Vienna (Universität Wien) to help digitize modern European exhibition catalogues. I used a combination of OCR and named entity recognition via RNNs to automatically extract and characterize information from digital copies of catalogues, such as artist names and exhibit titles. The group’s work was featured in The Routledge Companion to Digital Humanities and Art History.

GAPS Dark Matter Experiment

As an undergraduate, I was a research assistant for Professor Charles Hailey in the Columbia Astrophysics Laboratory. I helped fabricate and test lithium-drifted silicon x-ray detectors for the GAPS experiment.


Contact ✉️

walid [at] reverielabs.com