Shengye Tao
Undergraduate in Mathematics & AI · BIMSA Mathematics & AI training program
Beijing, China · RanchoTao@gmail.com · GitHub
Profile
I am an undergraduate researcher and builder working across learning theory, quantitative trading, cognitive augmentation, and AI agents / research harnesses. I am particularly interested in turning open-ended research and decision problems into reproducible experiments, auditable workflows, and working systems.
Current Roles
Research Assistant Intern · Beijing Institute of Mathematical Sciences and Applications (BIMSA)
Sep 2026 – Present
Research on stochastic processes, machine learning, and data-driven modeling.
Youth Assistant · Renmin University of China–Westlake University Joint Institute for Future Humanity
Sep 2026 – Present
Supporting research, projects, and youth collaboration.
Research
Training-Stage Behavior and Interpretability of Language Models · 2026 – Present
Studying layer-level behavior, causal interventions, and cross-checkpoint stability during pretraining, with an emphasis on reproducible experiments and auditable evidence.
Data-Driven Identification of Heavy-Tailed Stochastic Processes · 2026 – Present
Studying random time changes, subordination, non-Gaussian heavy-tailed noise, and data-driven stochastic-process identification.
Selected Projects
Visual Deadline — A cognitive-augmentation system for long-horizon planning, deadlines, task pressure, and attention management. GitHub
RanchoAutoResearch — An AI agent / research-harness experiment spanning literature review, experiments, code, auditing, and writing. GitHub
Oppor-Radar — A multi-source information pipeline with LLM summarization for continuously discovering academic and industry opportunities. GitHub
MarketManifold — A quantitative research project for exploring market structure with correlations, distances, dimensionality reduction, and clustering. GitHub
Wayline — A short-cycle AI product prototype originating from the HiYouth Hackathon. GitHub
Selected Activities
- HiYouth Hackathon, Participant, Sep 2026
- Peking University Machine Learning Workshop, Participant, Jul 2026
- Tsinghua Qiuzhen / YMSC AI Summer School, Participant, Jun 2026
Technical Toolkit
Python · PyTorch · Git · Linux · LaTeX · LLM/agent workflows · experiment automation · data analysis