Hyesoo Kim
Developer Relations | Software Engineer | AI Enthusiast
About Me
Interest:
Education
B.S. Computer Science
Sookmyung Women's University | 2021 - present
- Exchange program at University of Bern (Switzerland) with a focus in computational algorithms
- Masters-level courses in Machine Learning, Big Data Analysis, Text as Data and Automated Content Analysis
Work Experience
Developer Relations
UpstageSan Francisco, CA, USAπΊπΈ
Upstage AI specializes in building AI-powered solutions like language models and document processing tools to enhance automation, productivity, and decision-making for businesses.
- β’Develops LLM-focused contents, including:
- β’Assists in organizing and participating in 2 hackathons and developer meetups, providing support by answering participant questions and gathering feedback on Upstage's Document AI and Multimodal LLM products
- β’Builds relationships between developers and U.S./Korean teams by providing market insights and use cases, ensuring Document AI and LLM products resonate with developers and meet real-world needs
Software Engineer
BoxLadderSeoul, South Koreaπ°π·
AI-powered platform that connects startups with global investors and experts
- β’Developed a new search engine by transitioning data to a NoSQL format using Elasticsearch, achieving a 3x increase in search speed. The engine is optimized for startups, providing efficient matching with relevant professionals and venture capitalists
- β’Engineered an automation system using OCR technology to convert startup IR documents into data, automating text extraction and data indexing processes, resulting in a 35% reduction in manual processing time
Projects
EPL Predictor
May-Jul 2024- β’Developed an EPL Prediction Model using an Elo rating system, optimizing parameters (K-factor, Home Field Advantage) for accurate match predictions
- β’Ran 1,000 simulations to forecast match outcomes and updated Elo ratings for all EPL teams
- β’Built an interactive visualization dashboard to display predicted points, league rankings, and probabilities for key outcomes
Chatbot Solution for Bell company's Tech Community
- β’Developed an LLM model that delivers instant solutions via a chatbot, reducing search time from 30 minutes to less than 10 seconds, and streamlining the process compared to manually searching Bell's "Tech Problem Community"
- β’Implemented the application using a RAG system structure, which enhanced the LLM model's ability to retrieve and generate precise answers with 10-20% higher accuracy than ChatGPT
B2B Recycling Solution for Kanton Aargau: Plastic Recycle Platform
- β’Addressed environmental challenges for company Kanton Aagrau by developing a platform to connect businesses in the recycling industry
- β’Algorithmically generated recycling cycles based on distance and plastic types when companies inquired about plastic materials, optimizing resource allocation
- β’Built a recommendation system, utilizing ElasticSearch and AWS for scalable backend support
Ottogi Food Product Discovery
Jul 2023- β’Developed an application for Ottogi (food co.) that enables consumers to instantly discover recipes by scanning product images while shopping, combined with OCR technology for text recognition
- β’Created a chatbot service utilizing GPT-4 and LLM models with prompt engineering skills, combined with a RAG system, allowing customers to easily obtain detailed product information and reviews from Ottogi's extensive data, which reduced search times by 50%
AWS Cloud Resume Challenge educational content
Aug 2023- β’Developed a personal resume website as part of educational content to teach university students about cloud platforms and automated deployment processes, while serving as a mentor to guide them through the learning experience
- β’Successfully created a website hosted on AWS, demonstrating key skills and concepts, which served as a practical learning tool for students
- β’Chose AWS and Terraform to provide students with hands-on experience in cloud infrastructure as code, addressing the need for understanding automated and scalable solutions
ScanScribe: Intelligent Document Deck (IR DECK)
Aug 2023- β’Engineered an automation system using OCR technology to convert startup IR documents into data, automating text extraction and data indexing processes
- β’Achieved a 35% reduction in manual processing time through automated document processing and data extraction
Technical Skills
Programming Languages
LLM & OCR
Cloud Platforms & DevOps
Databases & Data Management
Development Tools & Frameworks
Awards
Wrtn Technologies Inc 2nd Generative AI Ideathon, SMWU
BaselHack 2023 - Bell food group
β’ Created an OCR-powered system to automate text extraction from documents, improving 50% of search efficiency
β’ Created a custom search engine using ElasticSearch to streamline information discovery for startups and investors
AWS x khu-campustown summer hackathon
ICCAS 2023 - Digital Therapeutics for Children with ADHD
Issued by Euro-Korean Women Scientists and Engineers Association
Participated in EKC2023 conference in Munich, Germany
Certifications
AWS Certified Solutions Architect Associate
Issued Apr 2024
Stanford Continuing Studies TECH 152 Z/H - A Crash Course in AI
Issued Oct 2024
AWS Certified Cloud Practitioner
Issued Jul 2023