BCI Resume

Boston, MA · Open to relocate · Available immediately

EEG/BCI Research Scientist
Neural Signal Machine Learning · Scientific Software

Ph.D. in Computational Science specializing in reliable EEG/BCI decoding under session drift. I develop leakage-controlled cross-session evaluation, confidence-gated source-selection methods, drift diagnostics, and reproducible Python/R research software. My work spans motor-imagery EEG, domain adaptation, Riemannian methods, and sample-paced pseudo-online inference.

Download BCI Resume (PDF) Email Yiming LinkedIn

Current two-page resume · Updated Aug 2026

Core BCI Focus

  • Motor-imagery EEG decoding and cross-session evaluation
  • CSP/FBCSP, covariance features, tangent-space methods, and Euclidean alignment
  • Domain adaptation, transfer learning, nested validation, and bootstrap uncertainty
  • Python/R scientific software, reproducible pipelines, Linux, Docker, and Slurm/HPC

Selected Evidence

  • First-author Frontiers paper: accepted for publication, Aug 2026, Article 1895016
  • BCI systems: CrossDA, ShiftDx, and NeuroStream
  • EEG competition: Kaggle HMS Silver Medal, rank 98 of 2,767 teams (top 4%)
  • Industry: two years of IoT data engineering before doctoral research

Selected Publications and Manuscripts

  1. Shen, Y., and Degras, D. “A Confidence-Gated Source Selection Strategy for Cross-Session Transfer in Brain-Computer Interfaces.” Frontiers in Human Neuroscience, Brain-Computer Interfaces. Accepted for publication, Aug 2026. Article 1895016. doi:10.3389/fnhum.2026.1895016.
  2. Shen, Y. “Small Calibration Cohorts Do Not Reliably Select Motor-Imagery EEG Systems for New Participants.” Under review.

Selected BCI Research Software

CrossDA

CLI and Python API for MAP, DWP, MMP, and BDP cross-session experiments with auditable summary, detail, and session-role outputs.

ShiftDx

Offline fixed-reference MI-EEG drift analysis across five shift metrics and no-adaptation, domain-adaptation, and retraining conditions.

NeuroStream

Sample-paced pseudo-online motor-imagery replay with four pipelines, Euclidean alignment, and progressive inference visualization.