Engineer / Researcher / Founder
- Prospective PhD student, Human–AI Interaction · Fall 2027
- Founder & CEO, Usercase (EarningBirds Inc.)
- B.Eng. Civil + Mechanical, SKKU, summa cum laude
SANGWANKIM
I build AI agents that study people: they watch what people do, ask why, and notice when words and actions disagree.
- 01Builds the whole system. Arduino test rigs, a HoloLens app built alone, a production AI research platform.
- 02Tests it on real people. A 50-user live-product study, ~1,200 sessions, 100+ discovery interviews.
- 03Redesigns from evidence. Rebuilt the AI moderator when what users said diverged from what they did.
01AR construction guidance
Conceptual reconstruction · not original footage
A HoloLens app I built alone. It lays the design model over the real site so a worker can see where a precast part should be installed.
First-person view through the headset, with the guidance overlay in the field of view. Stage 1 of 5: Scan. A full-scale precast slab is captured as a 3D scan and reconstructed into a mesh.
06 / Academic profile
Sangwan Kim
Engineer / Researcher / Founder · Seoul, South Korea
06.1About
I trained as an engineer in two disciplines, Civil and Mechanical, at Sungkyunkwan University, graduating summa cum laude in 2022. As an undergraduate researcher in the Structural Health Monitoring Laboratory, I worked on computer-vision crack detection, 3D scanning of precast concrete, an AR installation-guidance app for Microsoft HoloLens, and a numerical–physical hybrid-simulation testbed.
Since then I have built software products and, since September 2025, EarningBirds (Usercase): an AI user-research platform that links behavioral analytics to contextual, AI-moderated interviews inside real products.
Across all of it, the recurring problem has been the distance between a model and the world: a design model and the slab that was actually built, an AR overlay and the place it drifts to, what a user says was hard and what the session shows. In graduate school I want to study how intelligent systems can notice that distance in human behavior and ask about it well.
06.2Education
Sungkyunkwan University (SKKU)
2016–2022Bachelor of Engineering. Double major: Civil Engineering and Mechanical Engineering.
GPA 4.03 / 4.50 (95.3 / 100) · Civil 4.15 · Mechanical 4.16
Graduated with Highest Honors (summa cum laude), 2022
06.3Research interests
How can AI agents study people faithfully: asking from what people actually do, and knowing when a simulated person can stand in for a real one?
- AI interviewers grounded in behavior. Follow-up questions built from what a person just did, not only from what they report; when to step in, and how to ask without leading. (Usercase)
- Simulating people with agents. Agents built from a person's own interview; measuring where they answer like the real person and where they drift. (Dalm · willu)
- Where it started: sensing the physical world. Comparing design with as-built structures through vision, 3D scans and AR: measuring the gap between a model and reality, and showing its error honestly. (HoloLens · crack detection)
Questions for graduate research
- How does the act of asking change the behavior an AI system is measuring?
- How should an agent reconcile a person's account with their own behavioral trace?
- How do registration error and drift shape trust in spatial guidance?
Human–AI Interaction · AI-moderated interviews · Generative agents · Simulations of people · Behavioral data vs. self-report · Human-in-the-loop systems · Spatial computing · Civil × Mechanical × AI
06.4Research experience
Structural Health Monitoring Laboratory, SKKU
2020–2022Undergraduate Researcher · Advisor: Prof. Sung Han Sim
Can a camera see a crack the way an inspector does?
- Built MATLAB pipelines combining image preprocessing, edge detection and adaptive thresholding.
- Compared detection outcomes quantitatively across varying imaging conditions rather than tuning for selected images.
- Tested 3D point-cloud depth data for crack characterization.
- Finding: Subtle cracks were difficult to distinguish reliably from surface geometry using depth alone.
Coupling a computer model to a moving specimen
- Built and operated a single-axis hybrid-simulation testbed based on prior research, connecting computer-based simulation with Arduino-controlled stepper motors, assembled circuitry and LabVIEW.
Putting the design model where the slab should go
- Captured 3D scans of full-scale precast concrete slabs and reconstructed meshes to assess deviation from design geometry.
- Independently built and tested a Unity application for Microsoft HoloLens using Blender models, marker-based registration and spatial tracking.
- Finding: The 3D visualization was effective, but tracking drift limited precision installation guidance.
06.5Systems & entrepreneurship
EarningBirds Inc. (Usercase) · Founder & CEO
Sep 2025–now- Founded EarningBirds and led a team of up to five; co-architected and built the platform with an embeddable web SDK, linking behavioral analytics to contextual AI-moderated interviews.
- Shipped task-based usability testing with configurable completion criteria, path analysis and session replay.
- Redesigned AI moderation to use neutral, behavior-grounded follow-ups when reported difficulty diverged from observed behavior.
- 50-user, three-step study in a live seller-admin product; AI interviews, analysis and reporting in one day versus about two weeks previously.
- 48 of 50 participants reached the target through different interaction paths.
- Findings informed a decision to retain a new filtering feature.
- ~1,200 pilot/test sessions; 100+ customer-discovery interviews.
- Early product evaluations with researchers at Lucid Software and JPMorgan.
On-chain arbitrage · Co-developer & strategy lead
2022–24- Co-developed and operated an automated arbitrage system across DEXs, triangular routes and mempool-informed opportunities.
- Worked with the CTO on execution-latency infrastructure.
- About $38,000 net profit from $7,500 starting capital.
Bluetags · Founder
2022–24- Built and launched a Web3 intelligence product aggregating X and Discord data.
- Reached thousands of users through community-led distribution.
Blueroom · Founder
2022–24- Built and released an Oculus Quest VR/NFT application with interactive content and NFT trading features.
- Closed beta with approximately 50 early users.
06.6Recent AI work
Built with AI coding agents. Each one tests a question about how people and AI agents work together, and each left a note.
Dalm
Open ↗Stanford's generative-agent interview method, adapted to Korean: an AI interviewer runs a life-story interview, then builds an agent that answers like you.
LearnedLong interviews stayed on protocol only when the script lived in code and the model decided just the follow-ups.
willu
Open ↗A voice interview builds your AI twin; twins go on simulated dates with other twins to suggest matches.
LearnedWhen one model wrote both people, simulated dates barely separated good pairs from poor ones (55% vs. 49%). That pushed the design toward an independent agent per person.
Pizza Panic
Open ↗A game for kids where the AI assistant is confidently wrong once, so they learn to check it.
LearnedThree versions differ only in the hidden cause, so a player improves by checking the AI, not by memorizing answers.
Ringo
Open ↗Real situations where only English works; AI characters adapt to the learner every turn.
LearnedOne example sentence in the prompt was copied by all four characters, and banning questions made them passive. Change one thing, then simulate.
Sori
Open ↗Picture books for Korean-heritage children: reading a line aloud opens the next scene, judged by live speech recognition.
LearnedRecognition that waited for a pause made readers stop mid-sentence; continuous transcription with lenient matching fixed it.
Field notes
- Example phrases in a prompt become the common prefix of every output. Give rules, not examples.
- Ask a model to vary its output and it repeats itself. Roll the dice in code and hand it one choice.
06.7Publications
No peer-reviewed publications yet. Research to date is documented below as project records, each with its method, my role and what it found.
06.8Honors & awards
| Presidential Scholarship | Korea Student Aid Foundation (KOSAF) | 2020–2021 |
| Outstanding Startup Award | KOSME Youth Startup Academy | 2022 |
| Commendation from the Mayor of Suwon | Urban Regeneration Startup Project | 2021 |
| Dean's List | Sungkyunkwan University | 2020 |
| Highest Honor Merit Scholarship | Sungkyunkwan University | 2019 |
06.9Selected programs
| 500 Global Founders Retreat Selected for 500 Global's founder-development retreat and startup-networking program in the U.S. | 2025 |
| Smilegate Orange Planet Selected for incubation, workspace and U.S. market-entry support. | 2024 |
| KODIT Start-up NEST, 16th Cohort Selected to the Korea Credit Guarantee Fund's startup acceleration program. | 2024 |
| KOSME Youth Startup Academy, 12th Cohort Selected founder and class representative; about $60,000 in startup-program support. | 2022 |
06.10Technical skills
- Programming
- TypeScript, JavaScript (primary); Python, MATLAB, C#
- AI & software systems
- Embeddable web SDKs; browser extensions; behavioral analytics; LLM-moderated interviews; external API integrations; natural-language analysis
- Spatial & experimental computing
- Unity, Microsoft HoloLens, Blender, 3D scanning and mesh processing, Arduino, LabVIEW