Leo Yao Email & Phone Number

Cambridge, Ma, Us

Cambridge, Ma, Us

Co-taught weekly recitation to reinforce lecture concepts, demonstrate problem solving methods. Hosted weekly office hours to help students with problem sets and general conceptual questions. Prepared notes as supplemental resources and graded exams.

Cambridge, Ma, Us

Wrote computational model using PyTorch to auto-discover periodic orbits in chaotic systems using a variational and optimizer based method.Extended model to various physical systems and training methods through a functional programming approach.Optimized hyperparameters of model to tradeoff computational power, validation efficiency, and solution precision.Parallelized and deployed model to MIT OpenMind cluster, finding thousands of double pendulum periodic orbit candidates.Analyzed precision, stability, and connectedness of discovered classes of double pendulum orbits near bifurcation points.Presented poster + contributed talk at IAIFI Summer Workshop 2024; accepted to NeurIPS Machine Learning for Physical Sciences workshop; journal paper (first author) in preparation.Serving on Early Career and Equity Committee; advocated for a more inclusive and welcoming space for younger researchers. Proposed initiatives to better involve and engage undergraduates in IAIFI events and community.

Miami, Florida, Us

Engineered modular data loading library to create market features; designed and wrote implementations of data caching interface to decrease model loading time.Built pipeline for data transformations using Pandas; created consolidated regression scripts and single model configuration file.Optimized, cross-validated, and tested models using Scikit-Learn; ran distributed jobs to evaluate models across symbols using market-informed scoring metrics; achieved PnL improvement of 9% on simulations.Carefully documented code and research output; efficiently transferred work product to colleagues at conclusion of internship.

Theoretically modeled nonlinear optical systems; explored second harmonic generation. Used combinatorial methods to analyze and prove hypotheses. Summarized results and created visualizations.Wrote computational models in Python and Julia (QuantumOptics.jl, DifferentialEquations.jl) to confirm predictions. Optimized and scaled simulations to support thousands of interacting particles.

Planned and taught courses for elementary and middle schoolers, with content adapted to age group.Directed a camp-wide Science Bowl tournament as a fun activity; led the final onstage round live.

Bachelor Of Science - Bs, Physics + Electrical Engineering And Computer Science

Dual Enrollment, Mathematics

High School Diploma