From Saigon. Training machines in Boston.

MS in Business Analytics (Data & Methods) @ Boston University Questrom · BS in Finance & Business Analytics + CS Minor · First-gen student

AI Research Assistant @ BU — Benchmarking LLM guardrail security across 5 open-weight models. Building a 3-layer defense pipeline tested against 30 adversarial attack scenarios for interview avatar systems.

Sales Enablement Intern @ SOPHiA GENETICS — Automating CRM pipeline management and building data visualizations for a genomics company that analyzes tumor data for 1,000+ hospitals worldwide.

Childcare Market Segmentation · UMAP KMeans PCA

Clustered 2,500+ U.S. counties into distinct childcare markets. Silhouette score 0.24 → 0.47 — UMAP cracked what KMeans and Ward's couldn't. Built for policymakers targeting subsidy allocation.

Predicting 30-Day Hospital Readmission · ClinicalBERT Keras PySpark GCP

Catches 70% of patients who'd be readmitted within 30 days. Multimodal neural network trained on 50K+ ICU admissions from MIMIC-IV. Tested 13 model variants across 3 encoders. Best AUC: 0.703. Built to save hospitals from the $17B annual cost of avoidable readmissions.

H-1B Visa Approval Predictor · Gradient Boosting SMOTENC ElasticNet

2.23M+ immigration records. Gradient Boosting AUC 0.949, 87.8% accuracy. Feature-engineered STEM classifications, wage bands, and employer dependency flags from raw DOL data.

Multi-Label ECG Classification · PyTorch TransformerECG ResNet1D

Benchmarked 5 deep learning architectures (CNN, Transformer, GNN, Wavelet) on 27,765 twelve-lead ECGs across 5 cardiac conditions. Best macro-AUC: 0.885 (TransformerECG). Published in NEJM Statistics in Data Science style.

Finance Projects — DCF models (ATKR, TSLA), equity pitch decks (ATKR, SNCY), commercial airlines industry analysis, cross-functional strategy project, and guerrilla marketing pitch (DedCool).

Power BI Dashboard — Grocery retail analytics dashboard built in Power BI with sales, ratings, and target tracking.

Boston Scientific · Health Economics & Market Access Intern Insights on elective procedure volume and reimbursement trends → delivered to the Urology VP → presented at board level. Built cost models from Medicare claims data for devices used in 30,000+ annual procedures.

VinaCapital ($4B AUM) · Macro & Fund Analytics Intern Built a customer profitability model using regression in Python. Created the Chief Economic Officer's Q2 webinar deck on Vietnam's GDP and currency performance using Bloomberg + CEIC data.

Letters of Recommendation — Boston Scientific, VinaCapital, and faculty recommendations.

ML & Deep Learning · Python · TensorFlow/Keras · PyTorch · scikit-learn · PySpark · HuggingFace Data & Cloud · SQL · GCP · Bloomberg Terminal · CEIC Data Visualization & BI · Tableau · Power BI · Salesforce Dev Tools · Git · Claude Code · Java

📸 Photographer · 🍜 Eater of everything · 🐔 Raised chickens, parrots, hamsters, dogs & cats growing up — retirement plan is a farm and I'm not kidding

Full Portfolio · LinkedIn · kaihungtran@outlook.com

Open to full-time roles in data analytics, data science, and business intelligence · Graduating Jan 2027 · OPT + STEM extension eligible