Skip to content
View oliviacai210's full-sized avatar

Block or report oliviacai210

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
oliviacai210/README.md

Hi, I'm Olivia Cai

M.S. Business Analytics (STEM), USC Marshall | B.A. International Business, UC San Diego (Cum Laude, Phi Beta Kappa, 3.9 GPA)

I build end-to-end analytical systems: predictive models, cloud data pipelines, and AI-powered tools that translate messy data into decisions that stick. My background spans data and AI consulting, mortgage document intelligence, ecommerce analytics, and graduate research across finance, healthcare, media, and operations.

Currently open to full-time roles in analytics, data science, and AI/product strategy starting January 2027.

Featured Projects

Project Domain Key Result Stack
RAG Document Intelligence Pipeline AI / LLM Semantic search and extraction across 200+ page mortgage documents with benchmarked retrieval quality Python, LlamaIndex, FAISS, Streamlit, OCR
YouTube Trending Analytics Pipeline Data Engineering Serverless medallion data lake on AWS at under $1/month operating cost, querying 50 KB per Athena scan AWS S3, Glue, Lambda, Athena, PySpark, Tableau
Do Words Move Markets? NLP / Finance MLP AUC 0.565 for volatility prediction across 59K tweets (2010 to 2023); null direction result matches EMH Python, VADER, LDA, spaCy, scikit-learn
Insurance Loss Modeling ML / Actuarial 1st place case competition. Top-decile lift 3.2x over base rate on 40K policies LightGBM, XGBoost, SHAP, Tweedie
Healthcare Risk Stratification ML / Healthcare 3.5x relative readmission risk identified; Monte Carlo revenue projection $134.6M to $140.4M (95% CI) PCA, k-means, statsmodels, scipy

Skills

Languages: Python, SQL, R
ML / Analytics: scikit-learn, LightGBM, XGBoost, SHAP, VADER, spaCy, NLTK, statsmodels, TensorFlow, Keras
Data Engineering: AWS (S3, Glue, Athena, Lambda), PySpark, Snowflake, Kafka, Flink, Spark, Docker, dbt
AI / LLM: LlamaIndex, FAISS, OpenAI API, Streamlit, OCR (Tesseract, PaddleOCR), PyMuPDF
BI / Viz: Tableau, Power BI, matplotlib, seaborn
Other: Git, Jupyter, Snowflake, DBeaver, NoSQL

Background

  • M.S. Business Analytics (STEM), USC Marshall School of Business (2025 to 2026)
  • B.A. International Business (Cum Laude, Phi Beta Kappa), UC San Diego (2024)
  • Data and AI Extern, Outamation: built RAG and OCR pipelines for mortgage document intelligence
  • Lead Business Intelligence Intern, Foros Finance
  • Incoming Data and Analytics Intern, FTI Consulting (summer 2026)
  • Founder, PixelPerfectZoom: $8,593 in sales, 1,384% YoY growth in shop visits

Let's connect: LinkedIn | Email

Pinned Loading

  1. rag-document-intelligence rag-document-intelligence Public

    Production-style RAG pipeline for mortgage document Q&A. LlamaIndex, FAISS, BM25, Mistral LLM, and Gradio UI. Benchmarked on 15-question mortgage blob test set with 81.8% digital-doc Recall@4.

    Jupyter Notebook

  2. youtube-trending-aws youtube-trending-aws Public

  3. do-words-move-markets do-words-move-markets Public

    NLP analysis of public-figure tweets and presidential speeches vs. S&P 500 returns and volatility (2010 to 2023). 59K records, 5 classifiers, walk-forward CV.

    Jupyter Notebook

  4. insurance-risk-modeling insurance-risk-modeling Public

    End-to-end loss-cost regression and claim classification on 40K auto insurance policies. Tweedie XGBoost + isotonic-calibrated LightGBM. 1st place case competition.

    Jupyter Notebook

  5. healthcare-risk-stratification healthcare-risk-stratification Public

    Patient risk stratification on a 100K synthetic EHR cohort. PCA, k-means, A/B testing, bootstrap and Monte Carlo forward revenue projection.

    HTML