Nai-H. Pan

· GitLab · Taiwan · Vancouver, Canada

Statements

  • I am a senior software engineer working across backend, frontend, and cloud systems, preparing to apply for graduate study as part of a transition toward ML training and model fine-tuning work.
  • My recent research interest is in robustness and evaluation in board-game AI, using Go and KataGo as a concrete testbed. In particular, I have been studying adversarial failure modes such as cyclic attack patterns and how well they transfer across stronger models.
    Write-up: Part 1 · Part 2 · Part 3
  • Longer term, I hope to study training and architectural approaches that improve model robustness.

Current Experience

Senior Software Engineer — Upgrade, Inc.

2021–Present

  • Developed automated deposit capabilities supporting weekly and monthly cadence.
    • Consumer deposit platform grew from early stage to $1.1B+ in total deposits and $100M+ in quarterly payment volume.
  • Implemented a pipeline to charge off delinquent deposit accounts.
  • Integrated with RouteOne to ingest indirect auto-loan applications across the United States.
    • Indirect auto lending program reached $1B+ in serviced portfolio within a year and 800+ active dealer partners.
  • Implemented a customer notification workflow for physical mail delivery.
  • Developed an evaluation framework for processing pre-qualification offers for auto loans.

Other Experience

  • Developed an ad inventory management system, including an authoring tool that allowed customers to design and manage ads.
  • Worked on on-call software specialized in routing alerts to the appropriate recipient during incident escalation.
  • Implemented customer-data synchronization into third-party marketing systems, including Marketo and Salesforce.
  • Designed and implemented ETL and deployment workflows for Salesforce-related systems.
  • Worked on distributed configuration systems backed by Cassandra, including multi-datacenter synchronization and related service/client layers.
  • Worked on enterprise authentication and identity integration, including federation and corporate sign-on workflows.

Education

Bachelor of Applied Science, Computer Engineering
Simon Fraser University

Skills

AI / LLM Tools: llama.cpp, ComfyUI

Local Models: Mistral, Mistral NeMo, Stable Diffusion, Z-Image-Turbo, ACE-Step, Wan2.2

Core Technical Foundations: Java, JavaScript, Kubernetes, AWS

{hπ}