Office Hours — Goro Yeh

Prerequisites — read first

You have a specific, technical question — not "how do I get into ML" or "what should I learn."

You have already tried to solve it yourself: searched documentation, read relevant papers, written some code.

You can state your problem in one or two sentences. If you can't, the problem isn't specific enough yet.

You are prepared to share your code or diagram at the start of the call — not explain verbally from scratch.

You understand I may say "I don't know" — and that's a valid answer that ends the call early.

⚠ This is not for you if: You want general career advice, resume review, "which company should I join," or a mentor who will check in on your progress weekly. For those needs, please read the blog posts first — most of those questions are already answered there.

Request a slot

~3 min to fill out

Name

Current role & background What do you do, and for how long? (1 sentence)

What is your specific technical question? Be precise. "How does BEV fusion work?" will be declined. "My camera-LiDAR projection is off by ~3px even after calibration — here's the code" will be accepted.

What have you already tried? List the docs you read, the approaches you attempted, and why they didn't work.

Topic area

What does a successful 15 minutes look like for you? What's the one thing you'd walk away knowing?

Code / notebook link Optional but strongly preferred. GitHub Gist, Colab, or Replit.

I review applications manually and respond within 48 hours. Submitting this form does not guarantee a slot — you'll get an email with either a Calendly link or a note explaining why this isn't a good fit right now.

✓

Application received.

Your question looks specific and well-scoped. I'll review it and send you a Calendly link within 48 hours to pick a time. Keep an eye on your inbox.

In the meantime, here are resources that may already answer your question:

ML Roadmap with runnable code examples Free 12-Week AV Perception Kit Field notes on goroyeh56.com

→

Not quite ready for a call.

Based on your application, I think the resources below will answer your question better than a 15-minute conversation could — and you'll learn more by working through them yourself first.

Come back once you've hit a specific wall that the docs and code don't address.

Start with the ML Roadmap — interactive code for every topic Download the free 12-week kit with Docker + dataset scripts Read field notes from someone still building

Questions about this decision? Reply to the email I'll send you.

HOW IT WORKS

Four steps from question to call.

Fill out the form

Write a specific question with context. Vague applications are declined — the form is the filter.

I review it

Within 48 hours, you get an email: either a Calendly link or a note with better resources.

Pick a time

15-minute slots, video only. Show up with your code or diagram open and ready to share.

We solve it

I'll either point you to the exact thing you need, or tell you honestly that I don't know. No filler.

15 Minutes.One Real Problem.