CVXPY Hackathon 2026 (February 21st)

Thanks to all the attendees of the first ever CVXPY Hackathon! We loved seeing the great projects you created. We will continue to work with you to merge these new features into CVXPY, CVXPYlayers, solver-benchmarks, and other packages.

Projects

Authors	Description	Link
Aditya Sengupta UC Santa Cruz	Modeling radiative-convective boundaries in exoplanet atmospheres using DNLP.
Arjun Chandran & Rustum Chhor UC Berkeley	End-to-end MVP to compare LP routing with token + VRAM + model-switching penalty against an LLM planner.	GitHub
Emi Soroka Stanford	Added broadcasting support to Convex.jl for vectorized operations like .+, .-, and element-wise comparisons.	GitHub
Jack Topper NASA Goddard	Reconstructing solar wind speed from spacecraft observations using CVXPY.
Logan Bell Stanford	Markowitz problems for solver-benchmarks.
Luca Hartmann UC San Diego	Comparison of cvxpylayers with SPO+ for differentiation through shortest path problems.
Matt Colwell & Sam Machen Jacobi	Irreducible infeasible set reporting for CVXPY.	GitHub
Michael Howes Stanford	Regression and knockoff statistics problems for solver-benchmarks.	GitHub
Michael Klamkin Georgia Tech	Automatic dualization for CVXPY.	GitHub
Revant Kasichainula & Alan Xiao Stanford	Benchmarking and debugging of the POC CVXPY Rust backend.
Rohit Makhija Virtue	Portfolio optimization problem with historical data for solver-benchmarks.
Saeid Tafazzol CooperSurgical	A light library to convert sympy and casadi symbolic expressions to cvxpy atoms.	GitHub
Tim Zheng Dalhousie	Nearest correlation matrix (SDP) problems for solver-benchmarks.	GitHub