I am a PhD candidate at The Ohio State University and an M.S. CS student at Georgia Tech, working at the intersection of machine learning, causal inference, and computational economics. My research builds production-oriented models and decision systems — from RAG-based LLM pipelines over financial filings to Transformer sequence models trained on parallel corpora.
I am actively seeking Summer / Fall 2026 internships as a Data Science, Machine Learning, or Applied Scientist intern. I bring rigorous statistical training, hands-on deep learning implementation experience, and the ability to translate complex data into clear, actionable insights.
🛠 Technical Skills
Languages & Frameworks
| Category | Tools |
|---|---|
| Languages | Python · R · SQL · Bash |
| ML / DL | PyTorch · scikit-learn · XGBoost · Hugging Face Transformers |
| LLM & NLP | LangChain · RAG pipelines · FAISS · OpenAI API · sentence-transformers |
| Causal Inference | EconML · DID · IV · RDD · Synthetic Control |
| Data Engineering | Pandas · NumPy · Spark · SQL · web scraping (SEC EDGAR, REST APIs) |
| Geospatial | Google Earth Engine · GeoPandas · Landsat 8 / Sentinel-2 |
| Statistics | Bayesian inference · forecasting · time-series · A/B testing |
🔥 News
- 2026.03 Actively seeking DS / ML / Applied Scientist internships — Summer / Fall 2026.
- 2026.06 Paper accepted at AAEA 2026: The Energy Cost of Artificial Intelligence.
📝 Research & Publications
The Energy Cost of Artificial Intelligence: Modeling AI-Driven Data Center Electricity Demand with LLM Predictions
Danbo Chen, Zijun Zhou (Meta Platform)
American Agricultural Economics Association (AAEA) Conference, 2026
- Built a RAG + LLM pipeline (LangChain · FAISS · GPT-4o-mini) over SEC EDGAR filings and ESG reports to extract firm-level data center investment signals for six hyperscalers.
- Forecasted AI electricity demand 2025–2030; projected 239–295 TWh by 2030 (13–17% CAGR from 118 TWh baseline).
- Constructed a Regional Power Stress Index (PSI) identifying >10 high-stress grid corridors from spatially concentrated AI compute.
- [Additional working paper or publication], [Authors], [Venue, Year]
💡 Projects
Global AI Data Center Siting: Spatial Econometrics and Strategic Firm Behavior
Danbo Chen (The Ohio State University)
Working Paper
- Identified five strategic siting archetypes among hyperscalers using a firm–determinant intensity matrix built from geospatial, policy, and infrastructure data.
- Modeled siting probability across North America, Western Europe, and Asia–Pacific using location characteristics (renewable access, policy corridors, connectivity, PUE).
- Key finding: >90% of AI compute concentrates in 12 metro regions; siting is driven by grid reliability and policy incentives rather than user proximity.
Python GeoPandas Google Earth Engine Sentinel-2 Spatial Econometrics S&P Capital IQ
Consumer Shopping Behaviour Analysis — 2026
Danbo Chen
Full-Stack Data Science Portfolio Project
- Built an interactive Plotly KPI dashboard and 14-section visual analysis (Sankey, Sunburst, radar charts, heatmaps) on 11,789 consumers.
- Trained and benchmarked RF / GBM / XGBoost / LightGBM classifiers with 5-fold CV and SHAP explainability; GBM regression for online spend prediction.
- Applied causal inference pipeline: structural path model (logit + OLS), bootstrap mediation (95% CI), and propensity-score matching (ATT) to isolate spending drivers.
Python scikit-learn XGBoost LightGBM SHAP Plotly Statsmodels K-Means PCA Propensity Score Matching
Neural Machine Translation: From LSTM to Transformer — A Bottom-Up Implementation Study
Danbo Chen
CS 7643 Deep Learning, Georgia Institute of Technology, Spring 2026
- Implemented LSTM, Seq2Seq with attention, Transformer encoder, and full Transformer from scratch in PyTorch on the Multi30k English↔German dataset.
- Full Transformer achieved val perplexity of 6.9 vs Seq2Seq’s 25.4 — 3.7× lower — while training 2× faster per epoch.
Python PyTorch Transformer Seq2Seq LSTM Attention Multi30k
📖 Education
- Ph.D., Applied Economics, The Ohio State University, Columbus, OH
- M.S., Computer Science, Georgia Institute of Technology