Skip to content
View yullieyang's full-sized avatar
1 follower · 3 following

Achievements

Achievement: Pair ExtraordinaireAchievement: YOLOAchievement: Pull Sharkx2Achievement: Quickdraw

Achievements

Achievement: Pair ExtraordinaireAchievement: YOLOAchievement: Pull Sharkx2Achievement: Quickdraw

Block or report yullieyang

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
yullieyang/README.md

Yullie Yang

Applied AI Builder & Quantitative Analyst

I build AI agents, workflow automations, and analytical tools — with attention to evaluation, reproducibility, structured outputs, and human review. My background is in quantitative analytics and model validation; I bring that same discipline to applied AI work.

📍 Based in Boston · relocating to the San Francisco Bay Area

Featured GenAI / LLM solutions

Agentic AI Evaluation Platform

Streamlit Case Review page showing evidence available to the agent, a deterministic baseline, run diagnostics, deterministic validation results, and a human-escalation decision for one case

Reviews analytical monitoring cases, retrieves supporting evidence, validates structured findings, and routes uncertain cases to human review.

Python · Pydantic · Anthropic SDK · Streamlit

View Code · View Output · Read Walkthrough

CardNews AI

5 by 2 thumbnail grid of a real rendered 10-slide card-news deck on the topic microplastics, editorial layout with serif headlines and numeral watermarks

Turns a user-provided topic into validated slide content and renders a reviewable ten-slide visual deck.

Node.js · Claude API · Puppeteer

View Code · View Output · Read Walkthrough

Quantitative foundation

  • Model validation & risk analytics — stress testing, scenario and sensitivity analysis, model monitoring, credit-risk analytics
  • Experimentation & metrics — A/B testing, power/MDE, CUPED, SRM checks, metric design
  • Analytical workflow automation — Python, SQL, reproducible pipelines

Currently a Quantitative Analyst on Risk Analytics at CoStar. Previously federal analytics and RAG/LLM evaluation at Guidehouse, and research tooling at the Federal Reserve Board.

Selected quantitative & applied AI work

Contact

📫 yullieyang@gmail.com · Portfolio · LinkedIn

Pinned Loading

  1. r-macro-trade-commodity-forecast r-macro-trade-commodity-forecast Public

    Reproducible R workflow: FRED macro/trade/commodity panel, auto.arima forecasts for net exports, real GDP, and WTI, FX pass-through regression.

    R

  2. llm-research-workflow-assistant llm-research-workflow-assistant Public

    Responsible AI workflow prototype for research QA, documentation, and human-in-the-loop review.

    Python

  3. cre_stress_test cre_stress_test Public

    Production-style CRE credit-risk modeling pipeline — Python package + R/auto.arima companion + SQLAlchemy persistence + Streamlit dashboard. FRED, Google Mobility, Boston Zoning. pytest + CI.

    HTML

  4. agentic-ai-evaluation-platform agentic-ai-evaluation-platform Public

    Applied research study of LLM-based QA agents on synthetic model-monitoring anomaly review: scenario-labelled synthetic data, deterministic baselines, schema-constrained agent + reviewer, calibrati…

    Python

  5. product-ab-experiment product-ab-experiment Public

    End-to-end A/B experimentation analysis: North Star metric, SRM check, power/MDE, CUPED, guardrails, segmentation, and ARIMA forecasting. Reproducible.

    Python