Greetings Everyone,
I would like to contribute to Update CUTEst.jl to the Optimization.jl Interface and Add to SciMLBenchmarks.Here is my personal information for consideration as requested:-

Full Legal Name:- Arnav Kapoor

CV:- Research_CV__IISER_Bhopal.pdf, Linkedin

Short Bio

I am a Computer Science sophomore at Indian Institute of Science, Bhopal with strong experience in systems programming, numerical computing, and scientific tooling. My work spans C++, Julia, Python, Docker, Mathematica and ROS Infrastructure with contributions to performance-critical libraries, hardware-integrated robotics systems, and AI research infrastructure.

I have contributed to deep learning and simulation projects including:

• Semantic segmentation pipelines using Jetson + ZED 2i+Open3D at IISER Bhopal, apart from leveraging NVIDIA Isaac Sim and ROS Infrastructure for Uni-tree Go1 model. Mentioning sonata, livox-viewer, issaacsim and ros-bridge
• Integration of config-driven training and benchmarking workflows for 3D deep learning models (PVCNN, RandLA-Net)
• Contributed an optimization to Julia Base by improving method error suggestions using type information from invoke expressions#58464
• Enhancements to networking modules in ns-3, including new device API support and memory-leak fixes.
• Automated Postman and GitHub Actions tooling for interacting with generative model APIs.

I have also been selected as an LFX mentee at Linux kernel, where I contributed to infrastructure reliability tools in the Linux kernel. I also undertook a research internship at the University of Guadalajara, working on hybrid quantum-classical optimization algorithms.
I focus on building reproducible, scalable infrastructure for research and developer productivity in scientific computing such as GLoBES.

Project Description

I am interested in updating CUTEst.jl to the Optimization.jl interface and integrating it into SciMLBenchmarks.jl because I see this as an essential step toward enabling scalable, reproducible benchmarking in scientific optimization. CUTEst [Gould et al., 2015] offers a standardized, well-curated suite of constrained and unconstrained nonlinear problems, but its current usage in Julia is isolated from the broader SciML optimization stack. This limits its role in evaluating newer solvers or hybrid techniques. My goal is to bridge this gap through a two-part approach: first, I will build a robust translation layer between NLPModels.jl and Optimization.jl, ensuring compatibility with differentiable objectives and constraint handling. Second, I will design benchmarking scripts that loop over all converted problems and optimizers in Optimization.jl, compute convergence and performance statistics, and visualize the results via SciMLBenchmarks.jl.

I’m drawn to this project because I value building tools that simplify rigorous experimentation. Unifying CUTEst.jl with Optimization.jl will help the community iterate faster, compare fairly, and innovate confidently.