DEMETER · Ph.D. research, University of Florida · Aug 2022 – Aug 2026
BRIEFING
When agricultural data files move between organizations, the receiving party cannot check whether a file still matches what its source recorded. Healthcare and finance solved this under regulatory mandates. Agriculture has no such mandate, so the solution has to work when participation is voluntary and earlier parties may be offline.
WHAT I BUILT
DEMETER, a platform where each organization records a file’s fingerprint and handoffs on a shared Hyperledger Fabric ledger, signed with its own certificate.
Two checks behind one query: whether a file matches what its source registered, and whether a file built from other files traces back to the registered inputs its maker declared.
Go chaincode on Hyperledger Fabric behind a Node.js/Express REST API, a two-organization network on Docker with TLS on every node, and a plain-English question box with keyword rules and a local LLM fallback (Phi-3 via Ollama).
KEY DECISIONS
Files stay off-chain. The ledger holds only fingerprints and linked records, so each organization keeps its data on its own systems and no single party owns the shared record.
Verify without calling back. A receiver checks a file’s origin and history from the ledger alone, without any earlier organization needing to be online.
Report gaps, never guess. If a transformation step was never declared, tracing stops there and reports the gap instead of guessing across it.
The AI reads but cannot write. The language model only picks a File ID out of a question. Every answer comes from the ledger lookup, and the model has no path to change the ledger.
The ledger holds origin, transformation, and integration records; the files themselves stay off-chain.
LOOT
Tested on Florida soil-carbon data against sample-ID matching: it flagged a lab file with one changed byte, rejected a declaration citing an unregistered file, and resolved a sample ID shared by two soil horizons. Sample-ID matching failed all 3 cases.
A cross-sectoral review in Precision Agriculture (2026) and three infrastructure papers under review.
Code stays private until the papers are published.
MAIN QUEST · SYSTEMS · ONGOING
Own Connected Car Services (CCS)
System Engineer, Solomon America (Duluth, GA) · System Engineer, Hyundai AutoEver America · Sep 2026 – present
BRIEFING
Connected Car Services (CCS) connects Hyundai, Kia, and Genesis vehicles in the U.S. and Canada to each brand’s services.
WHAT I OWN
Develop and operate CCS services on Kubernetes across vehicle factory, carrier network, vehicle-control, and remote-command data.
Diagnose production incidents to root cause in the platform’s Java/Spring services and ship the code fixes; find security weaknesses in the code and ship hardening changes.
Build and operate carrier integrations (Verizon, Bell) that provision and bill vehicle connectivity.
Operate five internal data platforms (Python/Django) and 10+ automation tools for CCS data-quality monitoring, analysis, correction, and reporting, and build new tools and features on them.
Define integration designs and scope estimates for new in-vehicle data products.
LOOT
Scope: Hyundai, Kia, and Genesis vehicles in 2 countries (U.S., Canada), with 2 carrier integrations (Verizon, Bell).
Five internal data platforms and 10+ automation tools in operation, with new tools built on them.
[REDACTED] Everything else. Ask me about it in an interview.
MISSION · AI · CLEARED
Build Vision Models for Fish Farms
AI Engineer, Billion21 · Dec 2019 – Aug 2022
BRIEFING
Fish farms set feed amount and timing by manual observation, which is labor-intensive, inconsistent, and does not scale across species and tanks.
WHAT I BUILT
Computer-vision models in Python that tracked whole-tank fish feeding behavior as input to an automated feeding system and estimated individual fish size for growth monitoring.
My part was the vision signals. Other teams built the feeding control logic, edge hardware, and feeder mechanisms that depended on them.
KEY DECISIONS
One model per scale. YOLO tracked feeding behavior across the whole tank; a second model estimated the size of individual fish.
Train for the farm, not the benchmark. Both models were tuned for farm imaging conditions rather than research-grade datasets.
SIDE QUEST
AI Consultant, Global BioAg Linkages (Jan – Dec 2020): planned AI/ML adoption for agriculture and aquaculture companies, assessing technical feasibility and market viability for each.
STAT
Papers: 10
10 papers in total, including those under review
SELECTED SCROLLS
Cho, Y., Yu, Z., Ampatzidis, Y., Nam, J.-S. Facilitating a Future Agricultural Data Ecosystem: A Cross-Sectoral Review of Blockchain Application in Data Sharing. Precision Agriculture, 27, 138 (2026).
Cho, Y. et al. DEMETER: An Integrated Trust Infrastructure for Shared Agricultural Data. Under review.
Cho, Y. et al. An Enabling Infrastructure Layer for Origin Verification and Transfer Tracing of Shared Agricultural Data Files. Under review.
Cho, Y. et al. An Infrastructure Layer for Tracing Transformed Agricultural Data Back to Its Sources. Under review.
Methods for Selecting Spores from Laver Farm Devices. KR102034354B1, filed 2017, granted 2019. A camera photographs spores on seeding nets, and a control center measures spore quantity from the images to accept or reseed each net. Google Patents
Blockchain-Based Laver Supply Data Management Method and Smart Laver Supply Chain Management System Using the Same. KR10-2024-0011399, filed 2024 (pending). Records laver seeding, farming, and product data on 3 ledger nodes, one per stage.
STAT
Journals Reviewed: 7
Peer reviewer for 7 journals · 2024 – present
Carbon Management (Taylor & Francis)
Discover Agriculture (Springer Nature)
Discover Applied Mathematics (Springer Nature)
Environment, Development and Sustainability (Springer Nature)
Humanities and Social Sciences Communications (Springer Nature)
Mitigation and Adaptation Strategies for Global Change (Springer Nature)
Sustainable Futures (Elsevier)
COMPANION
Tobby
The working name of the AI assistant I am building, named after my cat.
ORIGIN STORY
Training Arcs
University of FloridaAug 2022 – Aug 2026
Ph.D., Agricultural and Biological Engineering · GPA 3.95/4.0
Texas Tech UniversityAug 2015 – Aug 2022
B.S., Computer Science, Minor in Mathematics · GPA 3.8/4.0