Selected Projects

ArviZ: exploratory analysis of Bayesian models

Core contributor, governance member, and primary dev of Julia packages.

PollenClim

Building a consensus probabilistic model of paleoclimate from simulations of climate models and fossilized pollen data.

Pathfinder.jl

Accelerating Bayesian inference and early diagnostics for probabilistic models.

Working out and benchmarking efficient invertible transforms from unconstrained to constrained spaces for probabilistic programming. See Stan implementations for probability vectors and for orthogonal matrices. Papers in prep.

e3fp

3D-aware fingerprint representations of small molecules for machine learning.

JuliaManifolds

Co-author of ManifoldsBase.jl interface for defining manifolds and algorithms on those manifolds, along with implementations in Manifolds.jl. paper

ChainRules

Core contributor to tools for defining and testing AD rules across the Julia ecosystem.

Pathfinder Benchmarks

Experimenting with different uses of Pathfinder for accelerating Bayesian modeling workflows.

Calibrated confidence bands for ECDFs

Numerical experiments testing the calibration of confidence bands for ECDF plots.

Intro to Supervised learning

Interactive teaching resource on supervised learning, from linear regression to neural networks. Developed for the IntroML workshop series at the University of Tübingen.

ManifoldMeasures

Lightweight implementations of common manifold distributions for Bayesian inference.

Seth Axen