Codify Scaling

Sam Lin
3 min readOct 21, 2022


Even with good effort, a fast-growing company often suffers from its total becoming less than the sum of its parts. Good tech companies typically deal with this as a bug rather than “a corp feature”. So, how is Tesla scaling much better than other AI corps? And, how did Amazon scale cross-team collaboration better?

1001 N Shoreline Blvd, Mountain View, CA

Auto labeling

While many things have become smarter by software-defined from phones to cars, not all tasks can be done by software yet. But, machine learning is accelerating the transformation. Psst, which can be opportunities for tech startups. Even for tech companies, it is easier to quickly start by manual operations. Therefore from time to time, there are news to uncover some AI services are actually “powered by human”, aka concierge MVP. Which provides a concierge service to figure out what to build. In the same time, labeled data can be accumulated to be the oil for ML. As tech teams learn more, it’s easier for them to codify the tasks to scale much better.

For example, Tesla recently shared Auto Labeling can label 10k trips in 12 hrs in 2021. Which used to take 5M hrs of manual labeling in 2018. Fundamentally, Tesla is a tech company instead of a traditional car maker. So, what else may you “use the force” to scale better 🦾?

Tesla AI Day 2: Auto Labeling

Collaboration API

Collaborating across teams is exponentially complicated at scale. And therefore, most companies pay more for ranking officers & build a hierarchy to make it work. While that may work to a certain extent, it alone is seldom a case across many teams of “smart creatives” at scale. What could be a “software solution”?

Better Coordination Was the Wrong Answer” as the book: Working Backwards argues. In order to make Amazon as a place where builders can build at scale & speed, Jeff Bezos transformed the software teams to collaborate through Application Program Interfaces (API) instead of meetings & emails, aka human coordination interfaces. It takes more effort to define & document the API. Often, it’s even painful to adapt as the iteration is more confined. But when it is done right, teams are freed to build autonomously and to iterate faster on their own terms & paces. Fundamentally, Amazon is a tech company instead of a digital market place only.

Dilbert: do more with less meetings, micro-management or money?

Full Disclosure

The opinions stated here are my own, not those of my company. They are mostly extrapolations from public information. I don’t have insider knowledge of those companies, nor a whatever expert. Up to you to take it with a grain of salt or two 😉.



Sam Lin

A Taiwanese lives in Silicon Valley since 2014 & has random opinions on some stuff. The opinions shared here are my own, not those of companies I work for.