← All work

production / case study

Kasikorn Business-Technology Group (KBTG)

System Engineer Intern · April 2026 – June 2026 · Nonthaburi, Thailand

Evidence

Accomplished a cost-efficient prototype as measured by a projected 20–50 USD/month operating cost, by doing vendor advisory classification with a Gemini-grounded intelligence pipeline.

Constraint

Architected a modular stability-patch classification pipeline for IBM (AIX/VIOS, MQ, WebSphere) and Red Hat (RHEL, JBoss, OpenShift) across 20+ vendors using Python, Pydantic schemas, and Hexagonal Architecture under enterprise governance constraints.

Decision

Prototyped a cost-efficient Gemini-grounded intelligence pipeline (projected 20–50 USD/month) to classify stability patches from unstructured vendor advisories, integrating Claude and feeding into ServiceNow CMDB via SharePoint + Power Automate.

  • Architected a modular stability-patch classification pipeline for IBM (AIX/VIOS, MQ, WebSphere) and Red Hat (RHEL, JBoss, OpenShift) across 20+ vendors using Python, Pydantic schemas, and Hexagonal Architecture under enterprise governance constraints.
  • Prototyped a cost-efficient Gemini-grounded intelligence pipeline (projected 20–50 USD/month) to classify stability patches from unstructured vendor advisories, integrating Claude and feeding into ServiceNow CMDB via SharePoint + Power Automate.
  • Automated multi-vendor advisory ingestion into SharePoint with scheduled Python workflows and Power Automate orchestration; defined MVP scope and HIPER taxonomy across 6+ product families while navigating strict security policy restrictions.

Failure modes

    Result

    Automated multi-vendor advisory ingestion into SharePoint with scheduled Python workflows and Power Automate orchestration; defined MVP scope and HIPER taxonomy across 6+ product families while navigating strict security policy restrictions.