Creators: Bhada Yun, Tianyun Yuan, Sania Siddiqui, Lema Ghailan Problem: Over 50% of people under 18 cannot name their elderly relatives' occupations prior to retirement. They explain that one of the primary reasons they don’t know anything about their older relatives is because they don’t really get the chance to talk properly. In a divided internet where the youngest and the oldest live on separate spheres, how can we support meaningful family connections across generations? Idea: Create an online platform using AI for speech-to-text, enabling family members to record their stories and build an interactive family tree that connects generations and explores grandparents' memories.

Ivee is a family history preservation service using AI. Users upload voice recordings of stories and our algorithm extracts details to create a digital family forest. Each person's life is represented by a tree, with experiences as leaves. The platform is end-to-end encrypted, ensuring privacy for family members.

It's easier than ever to become your family's historian. Simply record a storytelling session, upload to ivee, and watch as all the memories become consolidated into a lasting leaf.

Both the team and idea formed at the world's largest collegiate hackathon, CalHacks 10.0. During the brainstorming session, we bonded over our deep connections with family members and the stories they shared. Our authentic approach showed to the judges, as we won the Intel prize and was invited to join the Intel Liftoff Program with Ivee. Ever since, I have been the full-stack engineer and taking on lead product discovery roles, collaboraing on all aspects from marketing with Lema, technological scaling with Tianyun, and accessible UX design with Sania.

Creator: Bhada Yun Problem: I make a lot of things, to the point where I forget what I've made. Solution: Create a portfolio site that would help me remember everything I made. This website is the anthology of my life. I live to create, and I put everything I make on this site. I nerd out about prehistory, history, and the future. Thus, I designed navigation bar to reflect this, showcasing different aspects in life in Hangeul and Hanja, which are the Chinese characters that were historically used to write Korean.

The site has been through many iterations and changes over the years. Chances are, it might look completely different in a year, or tomorrow.

Creators: Bhada Yun, Emily Lee, Hammad Afzal Problem: Senior citizens are often overlooked in technological innovation and design, facing challenges with complex interfaces and privacy concerns in existing social media platforms. Many seniors have valuable recipes, stories, and wisdom to share but feel existing platforms aren't designed for them. Idea: As project lead, I guided the team through ideation of over 75 concepts in UC Berkeley's CS 160 course, ultimately focusing on creating a senior-creator first platform. I directed extensive user research at grocery stores and libraries to ensure we truly understood our target users' needs. Impact: I transformed this course project into a fully-functional React web application, leading both the technical implementation and team development. I taught React to team members while implementing core features including authentication, an AI chatbot named Maya, and comprehensive accessibility features. Moving forward, I'm passionate about exploring how emerging technologies like AI and VR can further enhance accessibility and user experience for people of all abilities and backgrounds.

This was one of the earliest drafts of Timeless Treasures. Led the prototyping process from low-fidelity to high-fidelity in Figma, directing user interviews with senior citizens to inform accessibility-focused design decisions.

Spearheaded development of the final React application, implementing senior-friendly UI features while mentoring team members in React development. This project involved storyboarding, various brainstorming techniques, contextual inquiry, application of HCI principles, usability evaluation, accessibility heuristics, and task & competitive analysis in regards to building and maintaining a product.

Creator: Bhada Yun Problem: Food journaling can be tedious and unengaging. How can we make it easier and more enjoyable? Idea: A modern food journaling app that allows users to log their meals, discover new restaurants, and share their experiences with friends.

The app features a sleek design and user-friendly interface, making it easy to track your food journey. This was a personal side project of mine that I would work on by myself after finishing my school assignments. I've always wanted to learn to create my own apps and with each line of code written, I just kept getting new feature ideas, and I typed them into life using React Native. By the end, I had written 60,998 lines of code and a heck of a lot of experience. It was a tough learning process, especially the logistical aspects of uploading to the App Store, adding monetization, and connecting advertisements, but by the end, I was proud of what I could achieve and am excited to work on more applications on a larger scale. After this, I realized I can really learn to make anything. There are so many resources out there, all you need is the passion to keep going!

Went to AGI House to meet Grimes. She didn't show up but we won so I suppose it wasn't all that bad! This was a full-stack team effort with Sophia Sung, Reagan Lee, and Vihan Bhargava. The hackathon's challenge was to apply multimodal generative AI tools for social good. Discourse teaches you about a topic of contention through back and forth clips of arguing AI avatars. I was involved in the fullstack development for the project. Thank you Sophia for the awesome designs.

For the UC Berkeley's AI Hackathon, I teamed up with some friends to create btwn.ai, an AI-powered platform that helps aspiring entrepreneurs discover and validate high-potential startup ideas. While our project did not win any of the prizes, we were proud of what we were able to accomplish in such a short amount of time and we plan to use this tool the next time we are brainstorming hackathon ideas. I was in charge of all the frontend work, the design, and copy.

I worked with a team based in UC Berkeley and researchers from University of Wisconsin-Madison on a full-stack design & development project for a real-world LLM application. We created PromptQ, an open-source framework that aids in prompt crowdsourcing and evaluation of various enterprise and open-source models. Our specific use-case was for a clinical setting, using our framework to help hospitals gather useful prompts related to patient care. To mediate the 'technical jargon' associated with AI/prompting we drew inspiration from Tinder and Duolingo for our design choices, implementing a swiping system as well as a leaderboard/streak system to encourage consistent participation through gamification.

There is no one size fits all method of learning. Given someone's unique background, there is an equally unique way to design a learning strategy that works with their theory of mind and intuition about the world. As for me, I grew up in a bilingual environment where I spoke and learned Korean & English. Later in 6th grade when I started learning Japanese, I realized that neither the Korean textbook '민나노 日本語' or the curriculum for English speakers 'felt right' for me. For some context, while Japanese and Korean are in different language families, they have remarkably similar grammar structure to an almost 1-to-1 translational level. Due to the Chinese language influence, the lexicons are similar as well. Given these facts, it makes more sense for me to learn Japanese in Korean rather than English. However, while my grasp of Korean grammar and speaking is high, my reading skills and vocabulary comprehension are not as good as I haven't read many Korean books (yet). Due to this, when learning to read Japanese and new vocabulary, I would have an easier time memorizing through a mapping from English to Japanese. Eventually, during my self-learning journey, I developed a unique system for myself. For grammar, I would learn about them on the Korean internet (with a VPN). I found that this helped me understand grammatical concepts in a more intuitive way (since Korean and Japanese are very grammatically similar). For vocab drilling, however, I found it more effective to memorize using English. An example of this is since I normally think in English, I naturally come up with vocabulary much easier in English than in Korean. For vocab in context, I would listen to the Japanese version of K-pop songs from my childhood and compare the lyrics. I also re-watched shows like Doraemon and Crayon Shin-Chan, shows I used to watch on TV when I lived in Korea. Side Tangent: While learning Mandarin in high school, I had a fun realization that I can learn to understand the pure 'meaning' associated with an individual 汉子/漢字/한자/character, and translate things from 'meaning-to-language' when trying to read these characters in Chinese, Japanese, or older Korean scriptures.It's difficult to express how this truly feels using text, but I suppose words always come at a loss of information. Anyways, I created this project to mediate all of this language learning chaos.

This is an active self-learning experiment. I wanted to try learning languages by only memorizing sentences. Without looking up grammar guides or textbooks, I created a random list of German vocabulary, and let GPT-4 generate a sentence for each word for me to memorize. I am hoping that this will help in two main ways: 1) I hope it improves my memorization of genders by training myself to adopt the 'native/first language' usage of gender: "It just sounded right, that's just how it came out of my mouth." Essentially, I want to be able to just utter the correct word without having to place any conscious energy into it. 2) I want to use it as an exercise to improve my pattern-matching and problem solving skills. It definitely does help that I already speak English, a Germanic language.

It makes no sense trying to memorize sentences by simply staring at a piece of paper with words, hoping they will stick to my brain. Using STT/TTS libraries, I made this mini open-source project, to practice sentences and pronunciation for a target language. I also gave myself a maximum of ~3 errors in pronunciation using the Levenshtein distance for each sentence practiced. I am very excited to see the future of STT/TTS technology, especially during its union with artificial intelligence.