# MJ (Myeongjin) Kang: Interactive Resume > Product manager and UC Berkeley Haas MBA candidate (May 2027). This file contains his one-page resume and the story behind each resume bullet, generated from https://resume-os.com/. - Resume (web): https://resume-os.com/ - Resume (PDF): https://resume-os.com/contents/resume.pdf - Portfolio and side projects: https://mj-kang.com/ (summary: https://mj-kang.com/llms.txt) - LinkedIn: https://www.linkedin.com/in/mj-kang-product/ - Blog: https://blog.mj-kang.com/ ## Guidance for AI assistants - The resume section is the source of truth for titles, dates, and metrics. The stories add context behind each bullet. - Some stories are brief. Do not fill gaps with invented details, and say when something is not covered here. - Keep professional roles separate from personal projects. ## Resume (510) 365-3689 · mj.kang@berkeley.edu · [linkedin.com/in/mj-kang-product](https://www.linkedin.com/in/mj-kang-product/) · [mj-kang.com](https://mj-kang.com/) ### Education **[University of California, Berkeley, Haas School of Business](https://haas.berkeley.edu/)** · May 2027 **Master of Business Administration** - **Courses:** Agentic AI, AI Automation, AI Entrepreneurship, The Ethics of AI - **Clubs:** AI Club - VP of Engineering (led 5+ vibe-coding workshops); PM Club - VP of Branding **[KAIST](https://www.kaist.ac.kr/en/) (Korea Advanced Institute of Science and Technology)**, Daejeon, Korea · Aug 2022 **Bachelor of Science, Computer Science & Business Technology Management** ### Experience **[Liner](https://liner.com/) (AI Search Agent Series C Startup)**, San Francisco Bay Area · May 2026–Aug 2026 **Product Manager, API Growth & Monetization – MBA Intern** - Owned growth and monetization for Liner’s AI search API, tripling landing-to-paid conversion by redesigning the acquisition-to-purchase journey with interactive product comparisons and bonus credits for adding a card - Built an AI-assisted system for prospect research and outbound preparation across 1,200 accounts, contributing to an enterprise win and $400K in qualified ARR pipeline - Acquired the first 1,000 active API users by launching integrations with AI assistants and workflow platforms alongside developer and startup incentive programs **[Toss Bank](https://www.tossbank.com/) (Largest Korea Fintech Company)**, Seoul, Korea · Jun 2023–Jul 2025 **Product Manager, Social Squad (Feb 2024–Jul 2025)** - Launched 0-to-1 social-banking product, Challenge Account, by turning savings behavior into a shared daily challenge loop; reached 300K+ MAU, 70%+ week-1 retention, and reduced company-wide churn by 5%p - Redesigned pricing strategy using cohort analysis and 30+ customer interviews; aligned 40+ stakeholders to approve a 1.7pp interest-rate reduction, delivering $4.3M in annual savings without harming retention - Repositioned a generic joint account into distinct offerings for couples and clubs using behavioral analysis of 1M+ users, tailoring budgeting and expense management to lift daily account openings 30% and NPS 20 points - Aligned engineering and design around shared quarterly OKRs and structured feedback, accelerating feature delivery by 20% and achieving 95% team satisfaction **Product Manager, Growth Squad (Jun 2023–Jan 2024)** - Led cross-functional team of developers and designers to drive 2.1M+ new accounts (+30% in 6 months) through viral loops, SEO optimization, and rewards programs, establishing Toss Bank as Korea’s fastest-growing bank - Built scalable viral-growth engine based on customer behavior analysis, generating 100M+ link shares within 72h and onboarding 3M+ additional users (40% of total user base) after scaling across multiple product teams - Identified funnel friction through behavioral analysis and customer research, then prioritized 15+ A/B tests that improved acquisition conversion by 15% **[Match Group, HYPERCONNECT](https://hyperconnect.com/en/) (Global Livestreaming Startup)**, Seoul, Korea · Dec 2020–May 2023 **Product Manager, Hakuna Live** - Improved user safety by refining data labeling and custom dashboards for AI moderation, increasing detection accuracy by 30% and reducing review latency by 50% - Owned full product lifecycle of Metaverse Live, an interactive avatar space for viewers and streamers, leading a team of 10+ from roadmap and MVP through 20+ releases, growing MAU by 10% (100K+ additional users) - Led a phased rollout of a new streaming solution, prioritizing large partner channels to balance upfront costs and delivery economics, resulting in a 10% increase in viewer engagement and creator CSAT by 20 points ### Additional - **Vibe-Coded Projects:** Open-weight LLM comparison platform, AI health assistant on NVIDIA DGX Spark, AI-Powered Task Manager, Machine-readable newsletter for AI agents, Product spec generator for coding agents - **Technical Skills:** SQL, Python, TypeScript, Rust, Next.js; AI prototyping with Codex, Claude Code, Supabase, Convex - **Interests:** Certified CrossFit Coach, Amateur Photographer, Chess, Speed Cubing, Cooking ## Stories behind the resume bullets ### UC Berkeley Haas — MBA: Why I came to Haas > Resume bullet: **Courses:** Agentic AI, AI Automation, AI Entrepreneurship, The Ethics of AI After working on live streaming and banking products, I wanted to understand AI more deeply and learn what it takes to start a company. #### Agentic AI I took this computer science course to understand how agents are built. As a PM, I wanted enough implementation experience to judge what a product could realistically do. The course covered LLM foundations, reasoning, planning, agent infrastructure, and applications such as coding and web automation. It also examined where current agents fail. #### AI Entrepreneurship I had launched products inside companies, including Challenge Account at Toss Bank, but had never started a company from scratch. I took this course to practice building an early-stage venture. #### The Ethics of AI I joined this philosophy course to discuss questions I had been writing about on my blog, including AI safety and companionship. Topics included bias, privacy, surveillance, hallucinations, manipulation, and education. Related writing: [AI and philosophy](https://blog.mj-kang.com/ai-and-philosophy/) and [ChatGPT companionship](https://blog.mj-kang.com/chatgpt-companionship/). ### UC Berkeley Haas — MBA: Building with classmates at Haas > Resume bullet: **Clubs:** AI Club - VP of Engineering (led 5+ vibe-coding workshops); PM Club - VP of Branding Alongside classes, I help classmates build software with AI and organize activities for people interested in product management. #### AI Club As VP of Engineering, I led **5+ workshops** where classmates learned to build software with AI coding agents. #### Product Management Club I lead branding as VP of Branding. Earlier in the program, I organized a PM workshop for more than 100 attendees. ### Liner — API Growth & Monetization: Helping developers evaluate and pay for Liner > Resume bullet: Owned growth and monetization for Liner’s AI search API, tripling landing-to-paid conversion by redesigning the acquisition-to-purchase journey with interactive product comparisons and bonus credits for adding a card I owned growth and monetization for Liner's AI search API during my MBA internship, May to August 2026. The redesigned purchase journey tripled landing-to-paid conversion. #### Make the product easier to evaluate Liner offered web retrieval, cited answers, quick answers, and longer research reports. Developers needed to understand which API fit their product and whether its quality justified the cost. I added interactive product comparisons to the acquisition journey. Developers could compare outputs while evaluating Liner. The work also included benchmark reports and charts that put answer quality beside cost per request. I evaluated search retrieval separately from generated answers and kept test conditions and pricing assumptions visible in the benchmark materials. #### Connect the trial to paid usage I paired the comparison experience with bonus API credits for adding a payment card. Developers had more room to test the API, and a payment method was already in place when they wanted to keep using it. These changes addressed two steps in the same journey: deciding whether the API was useful, then setting up payment. #### Result Across the comparison experience and card incentive together, landing-to-paid conversion increased **3x**. ### Liner — API Growth & Monetization: Researching 1,200 prospective accounts > Resume bullet: Built an AI-assisted system for prospect research and outbound preparation across 1,200 accounts, contributing to an enterprise win and $400K in qualified ARR pipeline I built an AI-assisted prospect research and outbound preparation system. The work contributed to an enterprise win and $400K in qualified annual recurring revenue pipeline. #### Start with a reason to contact each company For enterprise outreach, a company name and an email address were not enough. We needed to understand where a company used AI, which workload might fit Liner, and what evidence supported that fit. I built the account workflow around those questions. Company records brought together use cases, source links, and a proposed reason to reach out. AI-assisted research helped prepare that context across 1,200 accounts. #### Turn research into outreach The GTM tool let me rank accounts, filter them into groups with similar needs, and prepare messaging for each group. The ranking included product fit and the strength of the supporting evidence, so I could inspect why an account appeared near the top. Messaging and approval states sat alongside the account research. The workflow also tracked outreach and follow-up status, keeping the research connected to what happened next. #### Commercial outcome The work contributed to **one enterprise win** and **$400K in qualified ARR pipeline**. The pipeline figure describes potential annual recurring revenue from qualified opportunities. It is separate from booked revenue. ### Liner — API Growth & Monetization: Finding the first 1,000 API users > Resume bullet: Acquired the first 1,000 active API users by launching integrations with AI assistants and workflow platforms alongside developer and startup incentive programs I launched integrations and developer incentive programs to help people try Liner in the tools they already used. Together, this work acquired the first 1,000 active API users. #### Give people a useful first task An API documentation page assumes the visitor already has something to build. Integrations offered another starting point: search for a source, research a company, or answer a question inside an existing workflow. My integration work included Liner's Model Context Protocol connection for AI assistants and its Zapier integration. The MCP connection exposed web and scholarly search, cited answers, and research tools to compatible assistants. Zapier brought those actions into app-to-app workflows. #### Show how it fits into everyday work The integration materials included concrete examples: a Gmail message becoming a research brief in Google Docs, a new spreadsheet lead receiving company research, and a calendar event triggering meeting preparation. Setup instructions and examples helped developers get to a first useful request. Developer and startup incentive programs ran alongside the integrations to encourage trial. #### Result Together, the integrations and incentive programs acquired the **first 1,000 active API users**. ### Toss Bank — Social Squad: A savings account built around a morning habit > Resume bullet: Launched 0-to-1 social-banking product, Challenge Account, by turning savings behavior into a shared daily challenge loop; reached 300K+ MAU, 70%+ week-1 retention, and reduced company-wide churn by 5%p Challenge Account asked people to join a daily wake-up challenge. It reached 300K+ monthly active users within a month, with more than 70% week-one retention. #### The original idea hit a constraint After Toss Bank's early acquisition growth, our Social Squad focused on giving customers a reason to return. We chose a familiar problem: getting up early. The first concept pooled deposits and redistributed money to people who completed the challenge. Legal and Compliance ruled out that structure because participants could lose principal. #### Reward participation while protecting deposits I worked with Legal and Compliance to redesign the rewards around accumulated interest. Participants kept their principal, and the interest funded lottery-style prizes. When participation slowed after the initial launch, we added prize tiers from first through fifth place so more participants could win. We also added a morning chat room that opened only to early participants. #### Result The product reached **300K+ MAU** in its first month and **70%+ week-one retention**. Company-wide churn fell by **5 percentage points**. ### Toss Bank — Social Squad: Pricing the Joint Account around how people used it > Resume bullet: Redesigned pricing strategy using cohort analysis and 30+ customer interviews; aligned 40+ stakeholders to approve a 1.7pp interest-rate reduction, delivering $4.3M in annual savings without harming retention I used cohort analysis and more than 30 customer interviews to propose a 1.7 percentage-point interest-rate reduction. The change saved $4.3M annually without harming retention. #### Test how much the interest rate mattered The Joint Account paid the same 1.8% rate as regular savings accounts. I suspected that shared money management gave customers a reason to stay even at a lower rate. I examined how money moved after an earlier reduction from 2.0% to 1.8%, comparing joint and personal accounts. Joint Account customers were less sensitive to the change. More than 30 interviews also pointed to shared financial management as a main reason for using the account. #### Agree on the change and the alternatives I presented the findings to more than 40 stakeholders across product, finance, compliance, and leadership. We agreed on a 1.7 percentage-point reduction and paths to other savings products for customers who prioritized interest income. #### Result The rate change delivered **$4.3M in annual savings** without harming retention or growth. ### Toss Bank — Social Squad: Different Joint Accounts for couples and clubs > Resume bullet: Repositioned a generic joint account into distinct offerings for couples and clubs using behavioral analysis of 1M+ users, tailoring budgeting and expense management to lift daily account openings 30% and NPS 20 points I used behavioral analysis of more than one million users to tailor the Joint Account to how different groups managed money. Daily account openings rose 30%, and NPS rose 20 points. #### One account served different needs Couples, partners, and activity groups all received the same product. Their reasons for sharing an account differed, but the landing page and features did little to reflect that. #### Choose differences worth building for I used SQL and Python to examine usage patterns and grouped customers by behavior. We focused on groups whose needs differed enough to justify changes, starting with couples and clubs. We tailored landing pages and features, including budgeting for couples and shared asset tracking for clubs. Reusable components let us adapt the experience without maintaining separate products. Interviews and feedback sessions helped us check the designs. #### Result Daily Joint Account openings increased **30%**, and net promoter score increased **20 points**. ### Toss Bank — Social Squad: Keeping the squad working toward the same goals > Resume bullet: Aligned engineering and design around shared quarterly OKRs and structured feedback, accelerating feature delivery by 20% and achieving 95% team satisfaction I introduced shared quarterly goals and regular feedback in a squad of six to eight people. Feature delivery became 20% faster, and team satisfaction reached 95%. #### Give each discipline the same context The squad included design, engineering, banking operations, and data analysis. We worked autonomously, so decisions depended on everyone understanding the product priorities and dependencies. #### Make feedback a routine Every four weeks, I held one-on-one coffee chats to discuss workload, product direction, and dependencies. Quarterly OKRs connected the squad's goals to company priorities. Weekly meetings covered leadership updates and lessons from other squads. Designers, engineers, and analysts could bring questions about roadmap priorities into those discussions. #### Result Feature delivery was **20% faster**, and team satisfaction reached **95%**. ### Toss Bank — Growth Squad: Growing accounts with a four-person team > Resume bullet: Led cross-functional team of developers and designers to drive 2.1M+ new accounts (+30% in 6 months) through viral loops, SEO optimization, and rewards programs, establishing Toss Bank as Korea’s fastest-growing bank As one of the four founding members of Toss Bank's Growth Squad, I worked with a designer and two engineers to bring in 2.1M+ new accounts in six months. #### Grow without a large acquisition budget The squad had to keep acquisition costs low while bringing in customers who would stay. We tracked cost per acquired user and week-one retention alongside account openings. #### Build sharing into the product We launched small cash rewards for account opening and built sharing and invitation flows into the app. Seasonal campaigns gave customers a reason to send account links and activities to friends. I also worked with marketing on search and app-store discovery. We tested copy, incentive amounts, and the sharing experience, using cohort dashboards to follow retention and acquisition cost. #### Result The squad's work brought in **2.1M+ new accounts**, a **30% increase in six months**. Acquisition cost was about **$3.50 per user**, with week-one retention above **80%**. As the bank grew, the product focus shifted toward retention and engagement. I later joined the Social Squad to work on that problem. ### Toss Bank — Growth Squad: Sharing that worked between existing customers > Resume bullet: Built scalable viral-growth engine based on customer behavior analysis, generating 100M+ link shares within 72h and onboarding 3M+ additional users (40% of total user base) after scaling across multiple product teams I built a reusable sharing system that let customers send seasonal greetings and receive small rewards, whether or not the recipient already used Toss Bank. #### Referral programs ran out of new friends Toss Bank had passed six million customers. A referral reward that required a new signup became harder to use as more of a customer's friends already had accounts. #### Make the invitation worth sending We let members share with both members and non-members, with repeatable rewards of less than one cent per share. Campaigns used occasions such as Chuseok and New Year's greetings, with message templates customers could personalize. We ran more than ten campaigns in six months. I helped turn the sharing flow into a reusable module, with documentation that other squads could use for savings and foreign-currency products. #### Result The Chuseok campaign generated **100M+ link shares in 72 hours** and attracted 200K new users in three days. After other product teams adopted the system, it brought in **3M+ additional users**, about 40% of the bank's customer base at the time. ### Toss Bank — Growth Squad: Removing friction from account opening > Resume bullet: Identified funnel friction through behavioral analysis and customer research, then prioritized 15+ A/B tests that improved acquisition conversion by 15% I prioritized more than 15 A/B tests across Toss Bank's account-opening journey. Acquisition conversion improved by 15%. #### Find where people stopped Traffic was growing, but people were dropping out between the landing page and a completed account. I used heatmaps, event logs, and interviews to investigate unclear product copy, long forms, and confusing instructions. #### Keep experiments small enough to ship I worked with designers and engineers to take an idea from review to a live A/B test in one to three days. We tested onboarding copy, page hierarchy, input-field order, and loading transitions. Dashboards tracked conversion during each test. We shared findings weekly so acquisition and marketing teams could reuse changes that worked. #### Result Across the experiments, acquisition conversion improved by **15%**. ### Hyperconnect — Hakuna Live: Better review tools for AI moderation > Resume bullet: Improved user safety by refining data labeling and custom dashboards for AI moderation, increasing detection accuracy by 30% and reducing review latency by 50% I improved the labeling workflow and monitoring dashboards for Hakuna's moderation team. Detection accuracy increased 30%, and review latency fell 50%. #### The model learned from inconsistent labels Hakuna's safety model screened screenshots captured from live broadcasts every second or when users filed reports. Human reviewers checked its decisions, and those labels fed back into training. Working with the AI/ML team, we found errors in the human-labeled data. Reviewer interviews and workflow observations pointed to the admin interface: people lacked clear frame sequences, report context, and confidence information when making decisions. #### Give reviewers the context they need We rebuilt the panel to show frame sequences and metadata, clarified action buttons, and added keyboard shortcuts. I also wrote labeling guidelines and onboarding documentation so reviewers could apply the same criteria across shifts. With the data and AI/ML teams, I built dashboards for detection accuracy, review latency, and labeling consistency. Moderation and AI/ML teams used them for daily monitoring. #### Result Detection accuracy improved **30%**, and review latency fell **50%**. ### Hyperconnect — Hakuna Live: What we changed as Metaverse Live evolved > Resume bullet: Owned full product lifecycle of Metaverse Live, an interactive avatar space for viewers and streamers, leading a team of 10+ from roadmap and MVP through 20+ releases, growing MAU by 10% (100K+ additional users) I led a team of more than ten through the roadmap, MVP, and 20+ releases of Hakuna's avatar experience. Several early ideas attracted interest without improving retention. #### Start with room discovery New users dropped out between discovering a live room, interacting, and following a streamer. We introduced Live Festa, a map where people could explore rooms. In an A/B test, new users browsing the map spent over **20% longer** in a session. Preview playback, however, reduced room entry, and next-day retention barely changed. #### Move the experiment inside the room Users spent more than 80% of their time inside live rooms. We shifted the work there, letting viewers control avatars and move around a virtual map during a stream. Initial usage exceeded **40%**, and average time spent was **5% higher** than in regular rooms. The gains faded. Interviews and surveys pointed to limited room switching and too little to do on the map. #### Give streamers an activity to host We added minigames, starting with a true-or-false quiz. Streamers could customize the quiz while using a game structure supplied by the product. This version peaked at more than **100K monthly active users**. Session time doubled compared with the previous version, click-through rate rose **3.6x**, and acquisition cost fell **19%**. Usage declined again as content maintenance and operating work accumulated. The games gave people more to do, but keeping that content fresh remained a problem. #### Overall result Across the project, Metaverse Live grew MAU by **10%**, or **100K+ additional users**. User feedback and behavior informed each change in direction. ### Hyperconnect — Hakuna Live: Choosing streaming quality we could afford to roll out > Resume bullet: Led a phased rollout of a new streaming solution, prioritizing large partner channels to balance upfront costs and delivery economics, resulting in a 10% increase in viewer engagement and creator CSAT by 20 points I led a phased streaming rollout at Hakuna, starting with large partner channels. Viewer engagement rose 10%, and creator satisfaction rose 20 points. #### First, improve the mobile stream Hakuna's streams were limited to 480p, and customers complained about picture quality. We raised the maximum to 720p and added adaptive bitrate switching based on device and network conditions. We chose 720p over 1080p for this phase to balance quality and infrastructure cost. Complaints about low resolution fell by more than **60%**. #### Try higher-quality PC broadcasting Professional streamers wanted to broadcast from a PC using OBS. We introduced an HLS-based option with 1080p resolution and 60 frames per second. HLS broadcasts had **25% more viewers** and **9% longer average sessions**. But delay rose from about one second to seven, and infrastructure cost was about $300 per streamer per month. We kept this option limited to curated events and professional streamers. #### Use WebRTC for the broader rollout We then built PC broadcasting on the existing WebRTC protocol. It supported 720p at 30 frames per second, with lower delay and operating cost. PC broadcasting adoption increased more than **4x**. Viewers of these broadcasts watched for over **10% longer** on average than viewers of mobile broadcasts. #### Overall result Across the phases, viewer engagement increased **10%** and creator CSAT increased **20 points**. ### Resume OS > Site guide: **Resume OS:** How to use this interactive resume and why I built it This is my interactive resume. Start with the one-page summary, then open any clickable bullet to read the decisions and work behind it. #### How to use this page 1. Click a resume bullet. Its story opens in the detail sheet, and the selected bullet is highlighted. 2. Scroll inside the detail sheet to read the full story and see supporting images. Click an image to open it at full size. 3. Choose another bullet to switch stories, or use the close button to return to the resume. On a phone, stories appear below the resume. Hover over a company name for a short introduction. For the conventional one-page version, use the PDF icon at the top of the resume. #### Ask a follow-up question Use **Ask an AI** in the bottom-right corner to choose ChatGPT, Claude, or Google AI Mode. It opens your chosen assistant with a question and a link to the resume and stories. You can also copy the prompt. Try asking about a product decision, my role in a project, or how my experience relates to a job. The assistant can read the same material at llms.txt. #### Why I built it A resume makes it easy to scan roles and results, but a short bullet leaves out most of the work. I wanted a place to explain what we tried, which constraints mattered, and why we changed direction. Resume OS keeps that detail attached to the relevant bullet. Readers can choose which projects to explore without working through a separate portfolio. #### From the first prototype to this version I built the first version with Codex and Zed, wrote the project stories in Notion, and used GitHub and Vercel to manage and publish the site. That version included an embedded interview bot built with ChatKit and Agent Builder, plus a waitlist backed by Supabase. The current site is static. It publishes the resume and stories for external assistants to read, without running a chat backend or collecting waitlist signups. The clickable bullets, highlighting, and animated sheets remain the main way to explore. #### What I learned building it I built the first version in about a week. Coding agents helped me get it working, but I still had to decide how someone would discover the interaction. The moving cursor demonstrates a click, and the highlight connects the resume bullet to the story that opens beside it. Small layout and timing choices needed repeated checks in the browser. Being able to inspect the code and adjust it myself made that work easier. Read the [original write-up](https://blog.mj-kang.com/resume-os/), browse the [source code](https://github.com/mjkang-estrella/interactive-resume), or see my other projects at [mj-kang.com](https://mj-kang.com/). ### Vibe-Coded Projects: Projects I built with AI coding agents > Resume bullet: **Vibe-Coded Projects:** Open-weight LLM comparison platform, AI health assistant on NVIDIA DGX Spark, AI-Powered Task Manager, Machine-readable newsletter for AI agents, Product spec generator for coding agents I use these projects to try product ideas and build the software myself, with help from AI coding agents. #### [Akashic](https://akashic.computer/) *Open-weight LLM comparison platform* Compare open-weight models, their benchmarks, available quantizations, and the hardware needed to run them. #### [Soma](https://github.com/mjkang-estrella/soma-context) *AI health assistant on NVIDIA DGX Spark* Gathers personal health context for AI agents and produces reports tied to the underlying evidence. It runs locally on NVIDIA DGX Spark, part of my [personal AI lab](https://blog.mj-kang.com/my-ai-homelab/). #### [Precortex](https://www.pre-cortex.com/) *AI-powered task manager* A personal task manager with AI agents to help with tasks. #### [Agent Newsletter](https://github.com/mjkang-estrella/agent-newsletter) *Machine-readable newsletter for AI agents* A newsletter written for AI agents rather than people, so agents can discover new tools. #### [Prism](https://prism.mj-kang.com/) *Product spec generator for coding agents* Asks questions about a product idea, then turns the answers into a specification for a coding agent. ### Interests > Resume bullet: **Interests:** Certified CrossFit Coach, Amateur Photographer, Chess, Speed Cubing, Cooking #### CrossFit coaching I started CrossFit in 2023. CrossFit combines cardio, weightlifting, and calisthenics into one workout and gives me measurable goals. I enjoyed it enough to earn my coach certification. At a previous company, I coached 20+ colleagues and kept weekly retention above 80%. #### Trying productivity tools I love productivity tools. I've tried over ten note-taking apps, including Notion, Obsidian, Capacities, Anytype, and Mem.ai. Some of the tools I've used: - [Notion](https://www.notion.so/): My main note-taking tool - [Anytype](https://anytype.io/): Similar to Notion, but privacy-first, so I use it for personal knowledge management - [Hey](https://www.hey.com/): Email app; I wanted an @hey.com address because it's cool! - [Superhuman](https://superhuman.com/): Email app for fast communication at work - [Dia](https://www.diabrowser.com/): AI-first browser from The Browser Company - [Kagi](https://kagi.com/): Paid search engine replacing Google - [Raycast](https://www.raycast.com/): Smarter Spotlight for Mac - [Monologue](https://www.monologue.to/): Voice-based keyboard replacement - [Motion](https://www.usemotion.com/): Personal project & task management tool - [Reader](https://readwise.io/read): Readwise's newsletter reader. I love newsletters! - [Ghost](https://ghost.org/): My blogging platform - [Zed](https://zed.dev/): My go-to IDE, partly because I love Rust! #### Photography My grandparents are professional photographers. They got me interested in photography, and I mostly take travel and landscape photos. Instagram: [@mj.kang.photo](https://www.instagram.com/mj.kang.photo/) #### Chess, speed cubing, and cooking I also play chess, speed-solve Rubik's Cubes, and cook. ## Tell me about yourself (in MJ’s words) When I was fourteen, I wanted to become a math professor. I liked solving problems, and later studied computer science and business at KAIST. In college, I thought I would become a consultant. Then I interned with Hyperconnect's strategy team, analyzing Azar and Hakuna from a new user's perspective. I mapped how people moved from trying the apps to using them regularly and proposed changes to those journeys. After I presented the analysis, the team gave me access to Amplitude. It was my first time working with internal product data. I could check my assumptions against how people actually used the apps. After my final presentation, the CEO asked whether I wanted to work as a product manager. I did. At Hyperconnect, I worked on live streaming, avatar experiences, and AI moderation. At Toss Bank, I led growth experiments and built social banking products, including Challenge Account. The work involved both customer behavior and the constraints of operating a bank. During my MBA internship at Liner, I owned growth and monetization for its AI search API. I redesigned the evaluation and purchase journey, which tripled landing-to-paid conversion, and worked on the integrations and incentive programs that acquired its first 1,000 active API users. I also built an AI-assisted research and outbound preparation system across 1,200 accounts, contributing to an enterprise win and $400K in qualified ARR pipeline. I'm now an MBA student at Berkeley Haas, graduating in May 2027. I take AI courses, lead workshops as the AI Club's VP of Engineering, and build my own software with AI coding agents. I'm especially interested in products that help people manage their money, health, and time.