Tianlang (Jackie) Chen § 2 · EXPERIENCE
TIANLANG (JACKIE) CHEN · PERSONAL WEBSITE § 2 · EXPERIENCE
§ 2

Experience

5 RESEARCH · 1 INTERNSHIP · 2024–2026
R

Research

U-Michigan Research
P-2026-02 genomics in progress
Placeholder figure — EPCOT foundation-model pipeline (to be replaced) P-2026-02 FIG. 014

EPCOT Foundation Model: Data & Training Pipeline for TSS-to-Expression Prediction

Tianlang (Jackie) Chen

An end-to-end pipeline turning raw HPAP snMultiome data into per-donor, per-cell-type pseudobulk — ATAC single-cell 1 bp signal into 600 kb-windowed HDF5 arrays, and RNA counts into CP10K-normalized 19,264-gene vectors across 7 cell types. Training is sped up 2–5× per epoch at numerical parity via I/O prefetch, pinned H2D transfers, and vectorized masking; reset-head staged fine-tuning improves generalization on limited data; and a ~50 kb TSS-bin offset bug was found and fixed for cleaner interpretability.

↗ GitHub
U-Michigan Research
§3.01 research in progress
Placeholder figure (to be replaced) §3.01 FIG. 015

LLM for Incentive Design in Mobility Systems

Tianlang (Jackie) Chen

Investigating how large language models can support incentive-mechanism design for shared-mobility systems — ridesharing, transit incentives, and vehicle-sharing. Building LLM-guided workflows that generate and evaluate practical, interpretable incentive strategies, and studying how learning-based methods and optimization can be combined for data-driven transportation incentive design.

↗ GitHub
U-Michigan Research
§3.02 research in progress
Placeholder figure (to be replaced) §3.02 FIG. 016

AI-Guided Environmental Sensing & Adaptive Monitoring

Tianlang (Jackie) Chen

Developing AI-guided sensing for sparse, dynamic Great Lakes environmental systems under limited sensor coverage and seasonal variability. Builds reproducible workflows for data ingestion, baseline reconstruction, uncertainty quantification, and visualization, and compares random, grid-based, and uncertainty-aware sensor-placement and adaptive-sampling strategies — producing heatmaps, uncertainty maps, and placement recommendations for downstream deployment.

↗ GitHub
U-Michigan Research
§3.03 ml in progress
Placeholder figure (to be replaced) §3.03 FIG. 017

Functional Genomic Representation Learning for Microbial Communities

Tianlang (Jackie) Chen

A composable framework that turns microbial genomes into interpretable, parameter-free functional encodings via proteogenic k-mer tokenization, designed to plug into both classical statistics and large pretrained sequence models (Evo, ESM, ProtT5). Abundance-weighted community embeddings aggregate per-genome k-mer spectra into fixed-length community phenotypes, with leakage-safe feature engineering and honest statistical validation (permutation tests, careful cross-validation) for high-dimensional, low-sample genomic data.

↗ GitHub
U-Michigan Research
P-2026-03 webdev in progress
Placeholder figure (to be replaced) P-2026-03 FIG. 018

Web Applets for Data Analytics in Materials Science

Tianlang (Jackie) Chen

Interactive web applets (JavaScript front end, Python back end) for data analytics, curation, and sharing in the materials-research community. Users upload datasets and analyze them with state-of-the-art models, theories, and data-fitting methods; built with an existing team as reusable analytics modules and workflow templates aligned with FAIR data principles.

↗ GitHub
I

Internship

Mars Asia Internship
§3.04 systems complete
Placeholder figure (to be replaced) §3.04 FIG. 019

Full-Stack Development — Mars Asia (WeCom)

Tianlang (Jackie) Chen

Full-stack secondary development on WeCom (Enterprise WeChat) at Mars Asia's Global IT Service Center, using JavaScript, Node.js, and Vue.js to support internal enterprise communication and workflow tools. Contributed to API implementation, interface development, data migration, and information-retrieval features, and built automation workflows to keep operations running during a system migration.

↗ GitHub
FOL. 014 · EXPERIENCE
6 SPECIMENS · v1.0