Sewon Kim

Current

Innovation Accelerator Intern

Food and Agriculture Organization of the United Nations (FAO) · Office of Innovation

At FAO, I work on the ATIO Knowledge Base, the Agrifood Systems Technologies and Innovations Outlook knowledge platform. My responsibilities cover user engagement, behavioral analytics, search and UX evaluation, AI-enrichment quality, and knowledge-discovery workflows.

• Leads the user-engagement workstream and conducts behavioral analysis for the ATIO Knowledge Base, using GA4 to study how users discover and navigate agrifood innovation content.
• Analyzes search, device, entry-point, and profile-to-discovery behavior to identify UX friction and translate findings into product requirements.
• Evaluates AI-enrichment quality and contributes evidence for prompt, metadata, and knowledge-discovery improvements.
• Supports search, visualization, and taxonomy-aware discovery work, collaborating across UX, Drupal, taxonomy, data-harvesting, and AI workstreams.

Internal systems, non-public data, and restricted implementation details are intentionally omitted.

Undergraduate Researcher

Natural Language Learning Lab · Jeonbuk National University · Advisor: Prof. Hyun-je Song

Investigated knowledge-graph retrieval and computer-vision systems in a faculty-supervised lab, with responsibility for retrieval-pipeline, model-integration, and application-system implementation.

• Designed and implemented a retrieval pipeline and multi-hop reasoning mechanism that combined graph relations with neural embeddings using PyTorch and LangChain.
• Conducted ablation studies that isolated embedding-integration approaches to compare accuracy and computational-efficiency trade-offs.
• Implemented the Flutter field application, FastAPI backend, and DETR/CLIP detection pipeline for a collaborative cultural-heritage damage-assessment project.
• Documented research implementations, system architecture, and evaluation findings for lab and project materials.

International Scholar

IITP Korean AI Square Summer 2025 Program · Korean Software Square at Purdue University

Contributed field research and data collection to a government-funded international research project evaluating LoRa–MQTT monitoring of conditions inside and outside beehives.

• Supported data collection and field-test execution for a two-month trial using four sensor nodes across two beehives.
• Contributed to evaluating the collaborative system's LoRa transport, MQTT messaging, and sensor-data flow.
• Worked with the international research team to document the results and co-authored the six-author arXiv preprint as third author.