Skip to content
View ong6's full-sized avatar

Organizations

@NUS-Fintech-Society

Block or report ong6

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 is supported. This note will only be visible to you.
Report abuse

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

Report abuse
ong6/README.md

Ong Jun Xiong

Software Engineer at TikTok building AI infrastructure and backend platforms in Go and TypeScript.

I work on agent runtimes, tool servers, distributed services, data systems, and the interfaces engineers use to ship AI features safely. I built my team's shared AI platform from zero — an A2A + LangGraph runtime, a Go MCP server, and a React chat SDK — now in daily use by operations teams across the US, SEA and UK. Before the AI infrastructure role I spent 18 months on promotion and seller backends in Go and gRPC across 13+ markets.

Alongside the day job I ran Compoze, a one-person AI solutions company, on nights and weekends, and sold it to one of its clients in 2025.

How I build agent systems: tools and data feed an agent runtime, whose output passes a deterministic ground plane before reaching an interface.

Personal site · LinkedIn · Email

Selected work

A Python decision system that combines official transaction data, algorithmic scoring, and AI-assisted qualitative research for Singapore condo analysis.

Pythondata pipelinesalgorithmic scoringAI agents

My terminal-inspired recruiter site, with selected work, a browser and print-friendly resume, and an archive of university projects.

Next.jsReactChakra UIVercel

Current technical focus

  • Production agent systems using A2A, LangGraph, MCP, and deterministic evaluation boundaries
  • Go and gRPC services over event streams, caches, and analytical data systems
  • React and TypeScript interfaces for complex operational workflows

Most of my production work is proprietary. The public repositories above show how I structure systems, document decisions, and turn ambiguous personal problems into working software.

Currently building

Pinned Loading

  1. sg-property-analysissg-property-analysisPublic

    Singapore condo decision system — URA transaction data, algorithmic scoring, and AI-agent qualitative research into one comparable score per property.

    Python

  2. groundplanegroundplanePublic

    Deterministic fact boundaries for agent output — fails the build when a model's structured output asserts something its tools never produced.

    Python