At GlobalTek BPO Inc., I built a production multi-agent Disability-Appeal Strategy System using Python, Strands, AWS Step Functions, and DynamoDB. Its diff-based execution optimization scaled concurrent processing from 10 to 50 executions.
I also shipped an agentic document-processing pipeline with AWS Bedrock AgentCore, memory-backed agents, prompt-injection guardrails, and event-driven ingestion. It reduced manual processing by 130+ hours per month and cut error rates from 9% to 2%.
My work at GlobalTek also includes LLM evaluation and decisioning, AI-driven email workflows, and a multi-agent PRD assistant. These systems reduced email-processing effort by 70% and PRD drafting and refinement time by 30–60%.
Earlier, as a QA / Application Engineering Intern / Team Lead at GlobalTek BPO Inc., I built Playwright end-to-end test suites, shipped TypeScript features, and contributed full-stack work across production applications. I also built Atlas, a Rust, Tauri, React, and TypeScript developer tool for repository analysis and onboarding.

