Skip to content
View tipaek's full-sized avatar

Block or report tipaek

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
tipaek/README.md

Hi, I'm Timothy Paek

Software Engineer @ Synchrony • Backend Systems • AI/ML Background


About

I'm a Software Engineer building backend systems for digital payments (tokenization, API platforms, partner integrations).

I focus on:

  • reducing operational friction
  • automating repetitive workflows
  • making systems easier to understand and use

Recently:

  • Built an API testing framework that reduced regression time from ~90 min → ~2 min
  • Led multi-client integrations across distributed payment systems
  • Improving internal developer tooling and documentation at scale

Before that, I published ML research on detecting LLM-generated code (Springer Nature), achieving 99% accuracy on large-scale datasets.


What I'm Interested In

  • AI-assisted developer workflows
  • Backend systems at scale
  • Developer tooling & automation
  • Practical applications of LLMs in production

Projects

CodeClimb

Full-stack NeetCode tracker designed to reduce friction in interview prep
Tech: React, TypeScript, Spring Boot, Postgres

  • Multi-list tracking, activity calendar, streaks, "Up Next" queue
  • OpenAPI contract-first design to prevent FE/BE drift
  • Built using an agentic dev workflow (OpenAI Codex)

Rankhwa

Manhwa discovery and rating platform
Tech: React, Spring Boot, PostgreSQL

  • Search, filtering, user ratings, custom lists
  • JWT + Google OAuth auth system
  • Full-stack system design and deployment

LLM Code Detection Research

First-author published research on detecting AI-generated Java code

  • 99% accuracy, 0.999 AUC (Springer Nature)
  • Created datasets with 76K+ files
  • Reduced feature dimensionality 1000x while improving performance

Tech Stack

Backend

  • Java, Spring Boot
  • REST APIs, Microservices
  • PostgreSQL

Tools & Infra

  • Postman (automation/testing)
  • HashiCorp Vault
  • AWS

Other

  • Python (ML, automation)
  • OpenAI APIs / LLM workflows

Contact


Notes

I like working on hard problems, especially where systems, tooling, and AI intersect.

Pinned Loading

  1. NestedBigramsResearch NestedBigramsResearch Public

    Code for "Detection of LLM-Generated Java Code Using Discretized Nested Bigrams" (arXiv:2502.15740). Achieves state-of-the-art performance in distinguishing human vs. LLM-written Java.

    Java 1

  2. LeetCode_Brainrot LeetCode_Brainrot Public

    🎧 AI-powered podcast that explains all NeetCode 150 problems in TED-Ed style. No code — just clear, intuitive algorithm explanations you can listen to on the go.

    Python 1

  3. Rankhwa Rankhwa Public

    Rankhwa — Full-stack manhwa/webtoon discovery and ratings platform. Built with React + TypeScript (Vercel) and Spring Boot + PostgreSQL (GCP), featuring search & filters, JWT + Google OAuth, per-us…

    TypeScript

  4. GPT-Java-Dataset GPT-Java-Dataset Public

    A dataset composed of Java source code and GPT altered code for classification training.

    Java 1 1

  5. GPT-Java-GCJ-Dataset GPT-Java-GCJ-Dataset Public

    A dataset for testing large-scale GPT 4o detection.

    Java 1 1

  6. UchimuraTribute UchimuraTribute Public

    A Kohei Uchimura tribute page in html/css/js.

    HTML