Antares Yuan

A star to ship by.

AI Product Manager·Seattle | Beijing · currently working at ByteDance as an AI PM intern·résumé ↓·简历 ↓·Building in publicPi-Shaped AI PMAI Agent Native0→1HCI ResearcherHCDE @ UDub

30-second snapshot

what I'm building next

for collaborators

Antares's Token Consumption

— tokens·— sessions·— days active·since —

all charts → live

Shipped

6

The collaborative AI Agent rooted in the Lark ecosystem.

HACKATHON PROJECTAI Agent PMDelivery Layer DeveloperCollaborationLarkLark AI Challenge for College Students

GitHub(in Chinese) ↗Lark AI Challenge(in Chinese) ↗

Agentic RAG Reference Benchmark for Trustworthiness Evaluation

HACKATHON PROJECTAI Agent PMBenchmark

GitHub ↗Run the benchmark ↗

Gmail Plus Plus

Gmail++ is a AI-powered email ranking platform that gives you a read-only reply queue on top of Gmail, with visible reasoning and reversible account-scoped preferences.

SIDE PROJECTAI BuilderXGBoostProductivityWebsite

Try it ↗GitHub ↗

WorthFly currently focuses on search, decision support, alert drafts, and external purchase handoff. Final fare, inventory, and ticketing still belong to the real booking channel.

SIDE PROJECTAI PM0→1Explainable AIWebsite

Demo ↗GitHub ↗

Online benchmark for evaluating the susceptibility of CUAs to UI dark patterns.

RESEARCH PROJECTHCIIUI 2025

2026-02-23 0 cited by 2 papers

Paper ↗DOI ↗GitHub ↗

Coze Multi-Agent Orchestration

Lower the barrier for AI-coding novices to complete complex tasks and improve orchestration reuse for experienced users, through workflows tailored to each development task.

HACKATHON PROJECTAI Agent PMMulti-AgentCoze

2026-07-01 0 Shipped workflow recommendations, a visual editor and a template marketplace. Top 3 at the challenge.

Now

6

YiTang Profile Matching Infra

ApplyMint is a AI-powered chrome extension that compiles the resume into JSON, import it once for visual editing, and automatically fill it in one click on the job application form

SIDE PROJECTJob ApplicationChrome Extension

Designing Healthier Pre-Sleep Phone Use Experience for Teens & Early Adults

RESEARCH PROJECT

Ongoing project about building a system about using LLM to auto fix the dark patterns on the websites.

RESEARCH PROJECT

Kaboo — Token Efficiency

Token-consumption analytics platform. Led the Token Efficiency Leaderboard and the efficiency analysis module, now covering thousands of internal users.

INTERN PROJECTAI PMOpen SourceObservability

ongoing 0 Covering thousands of internal users.

Next

3

From PM to Builder

Stop briefing engineers — become one. Use solo builder loops to solve real business problems end-to-end, not just spec them.

Career PivotAI Builder0→1

Make AI worth looking at

Most AI products inherit the model's flat default aesthetic. Explore distillation + skill composition to ship AI that's well-crafted, not just functional.

AI AestheticDistillationSkillsCraft

Close the product loop

Build operations + data analytics fluency so I own the full product lifecycle — not just the design-and-ship slice.

Product OpsData AnalysisFull-stack PM

Later

2

AI for people, not just AI people

The 1-3 year version of me: shipped AI products that ordinary people — not the early-adopter or AI-insider crowd — use daily to make their lives meaningfully better.

MissionMass-impact AIReal-world AI

Human-centered, agent-aware

Human-centered design stays the foundation. But in the agent era, AI is also a user — and most products still ignore that. The 1-3 year version of me: deep enough AI fluency to design for both humans and the agents acting on their behalf, so the AI actually helps.

Human-centered DesignAgent-friendlyHCI × AIPi-shaped

$ ./agent terminal type help · ↑↓ history

antares@local: ~

$

$ ls ~/skills /skills dotfiles for AI — what I built to multiply my own AI work

$ grep -i how_i_think.md principles · updated occasionally

/ 01

Evals before features.If you can't measure it, you're shipping vibes.

/ 02

The model is one input.Latency, UX, and trust often matter more than benchmark deltas.

/ 03

Demos lie. Production tells the truth.Watch the long tail.

/ 04

A great PM removes ambiguity. Models add it back.The job is balance.

$ ./reach-out open to:

Currently open to: founder conversations in AI infra & agents, AI PM roles at top product orgs, and contractor work on 0→1 features. Tell me what you're building.

emailchenjy4@uw.edu linkedinChenjie (Antares) Yuan github@AntaresYuan DribbblleAntares Yuan

$ /agents personal data, kept separate not used by Ask Antares

Personal is a separate roadmap. Ask Antares and the root /llms*.txt files use Work content only; this page is not part of that corpus.

GET /content/*.json

Raw structured data: profile.json, board.json, lens.json, contact.json. Use for typed access to specific fields.

Stable IDs. Cards have permanent display IDs (SHIP-01, NOW-01, …) — citable across conversations and durable across content edits.
Pre-rendered HTML. The home page ships with content baked into the initial HTML — readable without executing JavaScript.
Separate from Work AI. These Personal records are not indexed by ask-index.json or copied into /llms-full.txt.
Open content. All data lives at /content on GitHub. Free to scrape, fork, or quote.
Build your own. This whole setup is an open template — “Use this template” on GitHub, npm run reset-content, deploy. There’s a writeup on the why.

$ comments notes from readers via GitHub Discussions