Yueh Han Chen Email & Phone Number

New York, Ny, Us

New York, Ny, Us

Advisors: Prof. He He and Prof. Brenden Lake.Working on: 1) Quantifying novelty in LLM-generated output 2) Safety evaluation/enhancement in human-AI Agent Interaction 3) Safety systematic generalization benchmark for language models

Berkeley, Ca, Us

· Worked with Prof. Jacob Steinhardt to build a retrieval-augmented LLM system for automated forecasting, capable of near aggregated-human-level performance. (NeurIPS 2024: Approaching Human-Level Forecasting with Language Models)· Collaborated with Forecasting Research Institute to curate the largest cross-domain benchmark dataset for evaluating the performance of LLM and human forecasters.

Mountain View, California, Us

· Devised a zero-shot textual tone classification system embedded within the Mailchimp product, resulting in a 37.9% increase in accuracy and 7x faster inference times, significantly enhancing the experience for over 100,000 daily customers.· Utilized SQL and Python to analyze 150M+ data points, producing actionable visualizations that guided decision-making for 5+ product managers.· Partnered with 4 software engineers and product managers to formulate customers’ problems into manageable technical tasks.· Improved the accuracy of the monitoring system for the tone model by 13% by implementing a zero-shot model as ground truth.

San Francisco, Ca, Us

· Conducted comprehensive research to identify the optimal ML model for detecting anomalies in API calls.· Analyzed 7 Billion API data points with Python, PySpark, and SQL, and helped build a deep learning model that enhanced anomaly detection accuracy by ~6% compared to the previous methods.· Supported product strategy with a data breakdown, aiding 3+ managers and engineers in understanding customer API usage.

Pleasanton, California, Us

Berkeley, Ca, Us

Client: Workday (https://www.workday.com/)· Implemented an autoencoder using Keras that helped clients identify 5% more anomalous expenses compared to the previous methods.· Analyzed 300K+ transactions, optimizing resources and delivering actionable insights that could save 5% in budget

Berkeley, Ca, Us

Client: Creative Artist Agency (https://www.caa.com/)· Implemented KMeans and Random Forest from end to end, including data cleaning, feature engineering, model building, cross-validation, and maximizing model accuracy to match college athletes with the best sponsorship with 92% accuracy.· Conducted 3 statistical analyses with Scipy that helped the client understand important factors for the KPI.

Mountain View, California, Us

· Increased click-through rate by 10% by providing actionable insights extracted from 2 A/B UI experiment analyses.· Analyzed 2M+ data with SQL to provide insights that helped the engineering team prioritize features to build.· Built 3 Tableau dashboards and extracted insights from them that helped 5+ PMs understand new weekly user trends.· Worked and communicated frequently with colleagues from the product, marketing, and executive teams.

Taipei, Tw

· Applied Random Forest to rank variables’ impact on North Star metric, suggested 3 A/B test ideas, 2 adopted by CEO.· Analyzed 4 A/B tests using Python & SQL to extract actionable insights on the product, resulting in 5% increase in KPIs.· Improved the declining IOS conversion rate by 10% through in-depth analyses of metric changes.· Devised a YoY metric summary to inform 2 stakeholders and 10 employees of business growth.DOWN is a location-based dating app with 160k+ monthly active users.

Taipei , Taipei, Tw

· Achieved a 3% increase in VIP conversion by implementing a decision tree model with an accuracy of 86%.· Increased VIP retention by 10% by designing and conducting experiments on incentive programs for first-time VIP users.· Built 6 dashboards using SQL from 1M+ customer data, which helped 6+ PMs understand weekly new customer trends.· Collaborated with the product, marketing, and design team and reported to the Operations team every week.【 𝘼𝙍𝙏𝙄𝘾𝙇𝙀 】What I Learned From my Internship at FunNow ➜ https://towardsdatascience.com/what-i-learned-from-my-data-analyst-internship-at-a-series-a-startup-ab8ef16b6eb1#9f10-9d82608fd1d3FunNow is a fast-growing Series-B startup that provides instant booking for leisure and entertainment activities in various Asia countries, including Taiwan, Japan, Hong Kong, and Malaysia.

Berkeley, Us

• Optimized a client's Yelp searching result from 3rd place to 2nd place while searching certain keywords.• Researched client's industry and gave the client a deeper understanding of the overall market.• Formulated marketing ideas to grow the client's social media.• Instructed fellow consultants on how to create effective visual representations.Outhrive is a pro-bono consulting service that connects Bay Area small businesses with creative students and working professionals. (http://www.outhrive.org/)

Taipei, Tw

• Negotiated with 25+ startup teams and conducted due diligence for a $10k USD investment.• Conducted market research on several industries(Healthy Food, Blockchain Media, Dating App, etc).• Co-organized 2 startup events.Rookie Fund is a venture fund that every year selects a cohort of students in Taiwan and Xiamen, China, puts money in their hands, and charges them with the responsibility for investing in local startups.Billed as Asia’s first campus-based venture capital (VC) fund, Taipei-headquartered Rookie Fund was formerly known as 500 Rookies, a name that speaks to its roots in the Silicon Valley-based VC 500 Startups. The fund is backed by Rui Ma, a former 500 Startups Head of Global Fundraising, and Norman Chang, also a former 500 Startups stalwart, and is currently overseeing the final months of its third batch of students.

Master Of Science - Ms, Data Science

Bachelor'S Degree, Data Science

Startup School