Aarya Bookseller

Undergraduate Research Assistant @IBM

College Station, TX, US

Focus: optimization for ML and ML-economics (bilevel optimization for principal–agent models), self-supervised representation learning (SimCLR/MoCo), and large-scale experimentation.Bilevel optimization (principal–agent models): Designed and implemented gradient-based bilevel methods using implicit differentiation with Hessian–vector products and conjugate gradient (CG). Formulated stable inner/outer loops with learning-rate schedules and regularization to improve convergence and reproducibility.Large-scale experimentation: Built automated pipelines for sweeps, logging, and result aggregation (CSV/Parquet) with publication-quality visualizations; standardized experiment configs to make runs reproducible across devices and seeds.Self-supervised learning (SimCLR/MoCo): Implemented contrastive pipelines in PyTorch; trained on A100 GPUs with strong augmentation policies; tuned temperature, optimizer, and augmentation hyperparameters; demonstrated improved transfer on downstream classification tasks versus supervised baselines.Research writing & collaboration: Contributed to theoretical formulation, experiment design, and manuscript drafting; maintained clean, documented code and figures for paper submission.

Delhi Public School Surat

High School, Science

Texas A&M University

Bachelor's degree, Computer Science