%20Chen-0A66C2?style=flat&logo=linkedin&logoColor=white){kind=icon}
B.S.E. in Data Science at the University of Michigan, Ann Arbor. Working across LLM-guided incentive design, genomic foundation models, and AI-guided environmental sensing. Previously Mechanical Engineering at Shanghai Jiao Tong University.
2026 — present· Research, University of Michigan
LLM incentive design, genomic foundation models, environmental sensing, materials-science tooling, microbial representation learning.2025 — present· B.S.E. Data Science, University of Michigan, Ann Arbor
Machine Learning · Computer Graphics & Generative Models · Computational Linguistics · Algorithms & Data Structures · Probability & Statistics.2025.05· Teaching Assistant, General Physics2025.03· Teaching Assistant, English Academic Writing2025.01· MCM/ICM — Honorable Mention
COMAP Mathematical Contest in Modeling.2024 — 2025· Mars Asia — Global IT Service Center
WeCom full-stack development (JavaScript, Node.js, Vue.js): API and interface work, data migration, cross-platform retrieval, and automation that kept operations continuous through a system migration.2023 — 2025.05· Mechanical Engineering, Shanghai Jiao Tong University
Applied Linear Algebra · Dynamical Systems · Design & Manufacturing · Dynamics & Vibrations.…· To be continued · 敬請期待
LLM for Incentive Design in Mobility Systems — Can an LLM design the incentives that make shared mobility actually work?
An LLM-guided workflow that generates and evaluates interpretable incentive policies for ride-hailing, transit, and shared vehicles — pairing learning-based methods with classical optimization.
EPCOT Foundation Model — TSS-to-Expression Data Pipeline — Made a genomic foundation model 2–5× faster to train — and found a 50 kb bug that was quietly wrecking its interpretability.
Pseudobulk pipeline from raw snMultiome (1 bp sparse ATAC → 600 kb HDF5 windows; CP10K-normalized 19,264-gene RNA vectors across 7 cell types). Five bit-equivalent training optimizations for a 2–5× speedup at identical numerics. Reset-head fine-tuning for small-data generalization. Located and fixed a ~50 kb TSS-bin offset.
AI-Guided Environmental Sensing & Adaptive Monitoring — Where should you put the next sensor in Lake Michigan? Let uncertainty decide.
Reproducible ingestion → reconstruction → uncertainty-quantification workflow for the sparse, seasonal Great Lakes, comparing random / grid / uncertainty-aware sensor-placement rules.
Applets for Data Analytics in Materials Science — Turning materials-science analysis into something you can just... upload a CSV to.
JavaScript front end + Python back end serving the materials community: users upload datasets and analyze them with SOTA models and fitting methods, built as reusable FAIR-compliant modules.
Functional Genomic Representation Learning for Microbial Communities — A parameter-free way to turn a whole microbial community into one honest vector.
Proteogenic k-mer tokenization yields interpretable, parameter-free genome encodings that plug into both classical statistics and pretrained sequence models (Evo, ESM, ProtT5). Abundance-weighted community embeddings, leakage-safe features, permutation tests and honest cross-validation.
More steamers on the next cart · 敬請期待
In preparation.
| Languages | Python · C · C++ · JavaScript · SQL · R · MATLAB · Java · Bash |
| LLM & Agentic AI | Prompt Engineering · Structured Prompting · RAG · LLM Workflow Design · Agent-Guided Code Generation · Output Validation · Hallucination Mitigation |
| ML & Data | PyTorch · scikit-learn · Gaussian Processes · Foundation-Model Fine-Tuning |
| Web | Node.js · Vue.js · HTML · CSS |
| Data Stores | MySQL · SQLite · HDF5 |
| Tools | Git · Slurm / HPC · Blender · SolidWorks |
| Spoken | English · Mandarin · Cantonese |
- Website
- jackiectl.com — Dim Sum Parlor — a 3D teahouse you can walk through (coming soon)
- Google Scholar (coming soon)
A stack of steamers, reduced along the last dimension, is one Tianlang (Jackie) Chen.