This CV contains links to verifiable credentials. Live version with working links: panitan.k.uphasia.co/cv-ambassador
PANITAN KWANKAEW
ปณิธาน ขวัญแก้ว (Jéan), Chiang Mai, Thailand
Law Student, Chiang Mai University
Founder and Managing Director at Uphasia Artificial Intelligence Co., Ltd.
I have been running something close to a Claude Builder Club for a while, without calling it that.
Students work with me across three faculties at two universities, Chiang Mai University and KMITL, including
ten law students on a public service project. We build on Claude Code together, on shared AWS infrastructure I set up and keep running for the group.
What I already run
2026
ongoing
Multi-tenant Claude Code environment, Chiang Mai University
Claude Code on AWS, set up and maintained for the group, at my own cost
- Built shared multi-tenant AWS infrastructure running Claude Code so several teams can work in one environment: isolated accounts, scoped permissions, bind-mounted project trees and per-repository deploy credentials, so several teams share one machine without reaching each other's code.
- Onboarding is now one double-click, with account provisioning, one-click launchers for Windows and Mac, pre-written project context, and read-only pair sessions for teaching three people at once. Two people with no programming background were building on the day they joined. The permission model lets them experiment without being able to reach another team's code.
2026
ongoing
SOUL GOOD, public service project
Ten law students, Faculty of Law, Chiang Mai University
- Working with ten law students on agricultural burning, volatile chemical use and public legal awareness. The supporting analysis covers 48,764 fire detections across three seasons and found 94% of burning was forest rather than agricultural, contradicting the common framing.
2024
to 2026
Running programmes before this one
Accelerators and a national AI cohort
- KX Knowledge Exchange, KMUTT, 2024 to 2025. TECHBITE 5 alumnus: my company Uphasia was one of the 27 innovations KX launched from that cohort. I then helped facilitate two cohorts, TECHBITE 5 and TECHBITE X Ajinomoto.
- AIAT Super AI Engineer Season 5, 2025. House Lead and Scrum Master, leading a team of 24 on a national AI programme.
- Zero to One accelerator, 2026. Programme facilitator on an internship placement, running talent headhunting and the kick-off event.
What I build
Dec 2023
present
Founder and Managing Director, Uphasia AI
Enterprise software and medical technology
- Uphasia ERP. Custom builds and Odoo implementations for medium enterprises in cosmetics and food, and for enterprise to public company conversion.
- ThaiDIN. Digit-in-Noise hearing pre-screening, deployed at two major public hospitals, Rajavithi and Ramathibodi, with the Department of Medical Services. Runs under Government Data Center and Cloud governance, serving 500+ patients. link
- Speech Elephant. On-device Thai speech therapy, so no audio leaves the phone. TED Fund innovation funding secured while still an undergraduate, validated past TRL 4.
2025
to now
law.uphasia.co, exam preparation for Thai law students
Built with Claude Code, on Thai open-model APIs
- Compresses a semester of statute reading into about an hour. Full-text search across the Thai legal codes with practice papers and AI-assisted marking. My classmates used it through midterms: 4,087 page views in exam week, 281 practice exam sessions.
- Checked the marking model before relying on it. Typhoon, a Thai open LLM, still varied between identical runs at temperature 0, so grading was constrained in code rather than by prompt.
2026
research
Floodtir, Bangkok flood surveillance
Five-person team at KMITL. Placed at BDI Hackathon 2026 Open
- Estimates water level from a single CCTV frame using YOLO, SAM and monocular depth, so flood depth comes from cameras already mounted across the city rather than new sensors.
- Working around data scarcity with a digital twin. Real flood imagery with measured depth barely exists, so we take Helsinki's publicly available city digital twin, a detailed and realistic 3D model of the real city, and flood it in Blender using Claude. Because we set the water level ourselves, every generated image comes with exact known ground truth. This is the method my Sony certification covered, high-precision AI across many locations using digital twins: convolutional features cross the simulation-to-reality gap well enough that synthetic data with perfect labels improves generalisation.
current
Research Assistant, Cyber Law
Law and Technology Study Center, Faculty of Law, CMU, to Assoc. Prof. Dr. Thotsapon Tasanapan
- Reviewing the latest Cyber Law Act draft on its technical substance: what it would require of businesses, what it would cost, and where the gaps are. The role needs enough engineering to know what a clause demands of a real system and enough law to know whether it says what it means. That combination is why I went back for the second degree.
Education
- 2026 to now
- LL.B. Law, Chiang Mai University
English-proficiency portfolio admission
- completed
- B.Eng. Computer and Robotics Engineering, Bangkok University
First Class Honours, finished at 20
Certification
- valid to 2029
- Human Research Ethics, Faculty of Medicine Ramathibodi Hospital, Mahidol University. EC No. 010769039. link
- 2026
- Internship Completion, CMKL University, a Carnegie Mellon and KMITL joint institution. link
- 2025
- Super AI Engineer (Developer), Season 5, AIAT. link
- 2025
- IELTS Academic, band 7.0 (CEFR C1), British Council. Thai native speaker. link
- 2024
- AI Development Workshop, Sony Group Corporation. link
Recognition and memberships
Competitions: 18 entered, 7 podium finishes, 3 won
- 1st place, Crime Charges Analysis Hackathon, AIAT and BOL
- 1st place, Liver Fibrosis Severity Prediction, AIAT and Chulalongkorn
- 1st place, Demand Forecasting Hackathon, AIAT and AXONS
- Also Sony Sensing IoT Award; Top 4, SCB10X Typhoon Hackathon; Asia-Oceania semi-finalist, S-Booster, Cabinet Office of Japan
Professional memberships
- Permanent member, Technology Promotion Association (Thailand-Japan)
- Permanent member, Artificial Intelligence Association of Thailand
- Corporate member, AI Entrepreneur Association of Thailand, through Uphasia AI
What I would do with a club
The hardest part of getting a non-programmer building is the first hour, and I have
already removed it for the people I work with. The club would start by opening that environment to the
campus rather than by booking a workshop, and it scales from a handful of people to a cohort without
anyone having to do anything differently.
I first raised this with Ian Borders at Claude Connect Chiang Mai this year. The medical
work is where Claude for Life Sciences would matter most to us: the hearing screening running in
two public hospitals and the speech therapy app both sit in that space.
It should run across faculties rather than inside one. CMU has strong medicine, law and engineering
schools that almost never build together, and every project above already crosses those lines. I would
also keep teaching the limits alongside the tool: in medicine and law a confident wrong answer is the
dangerous failure, so sourcing and honest uncertainty belong in every session.