Hi, I'm Hannah.
.gif)
I'm human. I have taste. I have emotions. Proof of both: my tumblr at 16 years old.
Since then, my taste has definitely improved. Or, I at least use Pinterest now
I'm interested in the intersection of technology and aesthetics, especially around fashion and personalization.
I have a B.S. and M.S. in data science and have since worked for Cuyana, QVC, and The Chefs' Warehouse.
02 — AI-Native
Skills
Languages & Libraries
PythonSQL
NumPyPandas
scikit-learnPyTorch
XGBoostPySpark
ML & AI
LLMsRAG
AI AgentsDeep Learning
NLP
Statistics
Causal Inference
A/B TestingTime Series
statsmodels
Cloud & Infrastructure
AWS
GCP
Azure Databricks
Tableau
UX Design
FigmaJavascript
HTMLCSS
AI Projects
001
Fashion Suite
End-to-end multimodal AI systems to track new product drops, search items semantically, and fill closet app market gaps.
CLIP EmbeddingsChromaDBRAGSemantic SearchRec SystemsAWS
002
Scaled Language Learning Newsletter
AI agent pipeline to automate newsletter and social media content creation, scaling the language learning newsletter model end-to-end.
AI AgentsAutomationContent Generation
003
Streaming Data Companion
Real-time data ingestion and chat analytics for live streamers. Dashboard tracks KPIs and saves chat history to answer questions about and engage viewers.
Real-Time IngestionChat AnalyticsDashboard
03 — Data Driven
Projects
001
Platform Adoption Model
ML and causal inference models driving 5–10% regional adoption increases for an online B2B platform.
MLCausal Inference
002
Customer Segmentation & MDM
Improved dataset quality by 20% through MDM strategy and ETL pipeline contributions.
PipelineFeature Engineering
003
Sales Forecasting Dashboard
Facebook Prophet model forecasting monthly store sales for the next 6 months, surfaced in a live dashboard.
Time SeriesViz
004
Fuzzy Matching & Auto-Labeler
NLP-powered TF-IDF matching tool that eliminated hours of manual labeling per pipeline run.
NLPAutomation
005
Product Recommendation Engine
Graph-based collaborative filtering model boosting e-commerce product conversion by 2% at QVC.
Graph MLE-commerce
006
Marketing Attribution Model
Data-driven attribution saving 15% of marketing spend, with SQL/Tableau dashboard for daily tracking.
AttributionSQL
04 — Accomplishments
Experience
The Chefs' Warehouse
Data Scientist
Platform adoption modeling, customer segmentation, and MDM pipeline work. Drove 5–10% regional lift and 20% data quality improvement.
Dataracy
Data Scientist
Sales forecasting and NLP automation tools featured as flagship case studies in the company's client portfolio.
QVC
Data Scientist
Led recommendation engine development underpinning two strategic personalization initiatives, boosting conversion 2%.
Cuyana
Data Analyst / Scientist
Multi-touch attribution model saving 15% of marketing spend. 0.97-accuracy neural network predicting repeat purchase behavior.
NYU Shanghai
B.S. Economics
Focused on econometrics, statistics, and data-driven research methods.