Wen-Ting (Joseph) Wang
About Me
I'm a machine learning engineer with nearly ten years of hands-on experience and a Ph.D. in Statistics. I build end-to-end systems that combine solid modeling with reliable data and infrastructure — and make sure they work at scale. Lately, I've been working on agentic workflows, RAG systems, and spatiotemporal modeling, blending practical engineering with research. My current focus is on building statistical boosters around pretrained models — things like conformal prediction and low-rank adapters — so that complex systems become both more trustworthy and easier to reason about.
More recently, I've been focusing on blockchain infrastructure and on-chain markets—using statistics, ML, and systems thinking to study market structure, execution, and liquidity, and to reconstruct and audit the infrastructure behind prices from public on-chain data.
Professional Experience
AI Product Engineering
2023–presentEntity linking for news analytics; RAG + MCP Q&A for an energy client.
RAG/NLP
E-commerce
2018–2023Large-scale recommendation systems from scratch.
Recommendation Systems/Large Scale
Data Science & ML
2016–presentNLP and time-series modeling across industries.
NLP/Time Series
Research
OngoingBayesian diffusion models; interpretable geospatial ML.
Bayesian/Geospatial/Diffusion
Technical Expertise
Machine Learning
Deep Learning/NLP/Time Series/Recommendation Systems
Technologies
Python/PySpark/PyTorch/TensorFlow/AWS/Azure/FastAPI/FastMCP/Docker/Kubernetes/Solidity
Research Areas
Diffusion Models/Geospatial ML/Bayesian Methods
Publications
- Routing Frictions and Executable Liquidity in Fragmented Markets
W.-T. Wang · arXiv preprint, 2026
- Calibrated Predictive Distributions from Sample-Based Generators
W.-T. Wang, S. Tzeng, Y.-T. Fan, H.-C. Huang · arXiv preprint, 2026
- Causal Effects of Protocol-Fee Changes on Liquidity Provision in Automated Market Makers
W.-T. Wang · arXiv preprint, 2026
amm-lab · GitHub
- Reinforcement Learning for Execution under Dynamic Fees in a Closed-Loop DEX Simulator
W.-T. Wang · arXiv preprint, 2026
amm-lab · GitHub
- Cluster-Aware Conformal Calibration for Spatio-Temporal Distributional Prediction
G. Kim, C. Y. Lim, W.-T. Wang, H.-Y. Huang, W.-Y. Wu · arXiv preprint, 2026
- Spatial Adapter: Structured Spatial Decomposition and Closed-Form Covariance for Frozen Predictors
W.-T. Wang, W.-Y. Wu, H.-Y. Huang, X. C. Wang · arXiv preprint, 2026
- DeepKriging on the Global Data
H.-Y. Huang, W.-T. Wang, P.-H. Chiang, W.-Y. Wu · Japanese Journal of Statistics and Data Science, 2026
- Regularized Spatial Maximum Covariance Analysis
W.-T. Wang, H.-C. Huang · Environmetrics, 2018
- Regularized Principal Component Analysis for Spatial Data
W.-T. Wang, H.-C. Huang · Journal of Computational and Graphical Statistics, 2017