Sangmi Yun
Software Engineer with a background in Linguistics. Accomplished Certificate in Web development from University of Washington Bootcamp, and ML&NLP from DeepLearning.AI. Comfortable in working with large datasets and possesses a disciplined work ethic. Completed a project in a team of three using language datasets and JavaScript to create games related to language. Currently, learning AI in Python in AWS scholarship program. With efficient communication and time management, we delivered our project on time and were awarded Best UI/UX Design.
I like playing video Games 🎮 I play most FPS games 🔫 such as Overwatch, Apex, Valorant, and more! I also like playing healing games 🧘 such as Terraria, Minecraft, Animal Crossing, and some VR games.
Experience
Data Labeling Analyst
- Collaborate with project managers to assess and improve external vendor performance, ensuring the safety, quality, and accuracy of labeled data for seamless integration into Llama3, contributing to the enhancement of Large Language Models.
- Spearheaded comprehensive compliance measures by documenting data across 4+ project groups, building and fixing guidelines, analyzing error points, and delivering vendor feedback, yielding a substantial 25% increase in accuracy.
- Implemented advanced data quality assurance measures for AI responses, ensuring 99% accuracy, logical coherence, and strict adherence to predefined criteria, bolstering the reliability and trustworthiness of machine-generated outputs.
- Conducted data on extensive language datasets, including video, text, and code data, and utilized SQL to extract and analyze performance data, significantly contributing to the development and delivery of MetaAI model responses.
- Developed a thorough understanding of data collection and annotation guidelines, enhancing the quality of labeled data and performing qualitative error trend analysis, while also managing the project budget to optimize resource allocation and cost efficiency, ensuring high standards of data quality and performance.
Education Coordinator
- Collected data on 200+ students and school profiles, offering valuable insights to gauge tutoring program demand and identify expansion opportunities for student-focused products to hit the monthly target of the enrollment students.
- Optimized Excel for schedule and student program development, enhancing visual representation and streamlining program management, resulting in a significant reduction of dissatisfaction to 2%.
- Recognized as the 'Best Education Coordinator' in Washington for successfully managing enrollment inquiries and increasing enrollment by 25% by creating monthly ads with promotions.
Annotator
- Analyzed 50+ instances of textual entailment ambiguity to evaluate AI's language understanding capabilities.
- Classified sentences based on the criteria of three which are entailment, neutral, and contradiction.
- Defined ambiguity according to the Lexical, Syntactic, Semantic, Pragmatic, Figurative, and other linguistic aspects.
Student Assistant
- Responded to 700+ visitors per month for their needs by providing varying school and contact information.
- Assisted returning freshmen, and other students at the beginning of the quarters to get their student ID cards.
- Oversaw and distributed inventory, organized stock for employees and visitors, and ensure we have enough.
- Received publication requests from the school department or other organizations and prepare the package to send out.
- Mailed the information packets to the prospective student, according to their interest.
- Supported school events like ‘Admitted Student Preview’ to share the UW experience with the Freshman.
- Worked on 5+ projects such as making a publication about the information that we were frequently asked for.
- Posted on the University of Washington’s official social media accounts to give information about the school.
Portfolio
Flower Image Classifier
Python project using Machine Learning to build an AI. Code for an Flower image classifier built with PyTorch, then convert it into a command line application
Skills: Python, Numpy, pandas, Matplotlib, Linear Algebra, Calculus, Neural Networks, Pytorch
Github Repo
Dog Image Classifier
utilized the built-in ML system (VGG) to classify the breed of the dog and with the result, we can identify the image that it is dog or not.
skills: Python,Data Types and Operators, Data Structures, OOP, Scripting, Functions, Control Flow
Github Repo
Syntax Structure of L2 Learning
Influence of L1 word order on L2 Interpretation
Education
University of Pennsylvania
AWS AI & ML Scholarship Program
Stanford | DeepLearning.AI
University of Washington Bootcamp
Univerisity of Washington
GPA: 3.8
Skills
- Proficient in Praat, and Microsoft Office (Excel, Access, Word, PowerPoint). Multilingual (Fluent English, Native Korean, Intermediate Mandarin, Intermediate ASL, Novice Japanese)
Awards & Certifications
- Annual Dean's List 2020 - 2022
- 1st Best UI and UX design Award
- 1st Best presentation Award