Sarah Wan

AI Product Manager Intern

TikTok

Feb 2026 - Current (7 months)

• Global AI Visual Search Expansion: Drove the go-to-market strategy for TikTok AI Visual Search in 33 European Union countries; Guided 5 engineers on AI search requirements and localization strategy, successfully served 250M DAU • AI Search Experience Eval: Built an eval-framework for two AI search features on TikTok from 0 to 1; Conducted data analysis of 1.2k queries, analyzing 4 dimensions: visual accuracy, text accuracy, latency, and safety, improving the hillclimb benchmark by 10% • AI Search Competitor Analytics: Benchmarked Google AI Search Feature user intent classification system; conducted reverse engineering of the Amazon AI E-commerce search on the visual Try-On feature, and established a better current product position. • Larg

AI Product Engineer Intern

OpusClip

Oct 2025 - Dec 2025 (3 months)

• End-to-End User Research & Product Insight: Owned end-to-end creator research for Agent Opus, an AI video-editing agent launched Oct 2025; surveyed & analyzed 600 + users to define workflow and strategies for v2 roadmap • AI Feature Design Evaluation: Performed competitive teardowns of Runway and Bytedance CapCut agents and refined asset-sync logic while reducing render latency 18 %; developed golden-sample standards for internal evaluation

AI Data Engineer Intern

Anuttacon

Aug 2025 - Jan 2026 (6 months)

• Data Market Mapping: Benchmarked over 58 global data suppliers to identify high-fidelity datasets for emotional-response modeling. Designed the company’s first data-sourcing evaluation framework, improving supplier screening efficiency by 35% and expanding model-ready data coverage across 5 new modalities (LLM/Vision/Audio/Video/Multa) • Video Strategy Development: Spearheaded partnerships with 3 Hollywood post-production studios to access cinematic underwater footage previously unavailable in the market. Enhancing creative scene diversity by 15% and reducing acquisition costs by 30%

L

Research Assistant

Language Technologies Institute, Carnegie Mellon University

Dec 2024 - Current (1 year 9 months)

• Data Automation & Evaluation Science: Designed Python-based semi-automated RLHF annotation workflow to benchmark LLM responses across dialogue contexts; boosted evaluation throughput 4× and reduced manual variance by 35% • LLM-as-a-Judge Evaluation: Developed evaluation benchmark for 2.4K UX queries to identify bias clusters in model feedback • Visualization & Research Communication: Produced data storytelling dashboards (ridge plots, correlation heatmaps) to guide model team’s evaluation roadmap

T

Founding Product Lead

Thetawave.ai

Jun 2022 - Feb 2025 (2 years 9 months)

• End-to-End Prototyping & Research Launched SmartPrep, an AI learning agent for test prep; translated learning-path data into Figma flows and prompt structures that cut time-to-solution by 37% and increased weekly active users by 4.2× • AI Evaluation & Model Design Partnered with engineering to benchmark GPT-based tutoring agents on accuracy, latency, and cost trade-offs; devised rubric scoring and HLE-style evaluation to guide LLM model selection for production use • Strategic Partnership Development Secured collaborations with TOEFL prep centers through demo negotiations; expanded the B2B pilot pipeline by 65% and established a data-sharing agreement supporting model fine-tuning efforts • Product Strategy & Impact Built financial project

About

Detail

Timeline

Résumé