I talked with Shaundai Person about privacy UX at Netflix, the micro-sales habits she brought from a decade in sales, and why AI should supplement engineering judgment instead of replacing it.
Shaundai and I go way back from the Twitter and conference circuit. These days she is a senior software engineer at Netflix, full stack on privacy and consent for Netflix.com. That work is easy to underestimate if you think "privacy page" means a legal footer and a checkbox. What she described was much more product-shaped: the right experience for the right person, under different laws, profiles, countries, and account relationships - including making sure you are talking to the adult on the account, not the kid's profile.
Put yourself in the user's seat first
When Shaundai pushes back on a feature, it is rarely a big disagreement. She walks the flow as a user. What if my kid picked this up? What if I was still on an ex's account? Then she brings that scenario to design, product, or marketing as a question instead of a lecture.
That habit matters more than the privacy domain. The best product engineers I know do the same move: experience the product as a human, then ask the team what they think about the rough edge they found. Sometimes the team already accepted the trade-off. Sometimes it is news. Either way, you earned the conversation by doing the work of empathy first.
Micro-sales, not prescriptions
Shaundai spent more than ten years in sales before she became an engineer. You can hear it in how she collaborates.
She does not walk into a service owner's Slack thread with "I need you to do this." She sells a series of smaller yeses: talk through the problem with me, help with this piece, teach me how you solved step one so I can do it next time. She shows up with a goal, a map, a guess at where the other person fits, and a real invitation for their dissent - because they own that service and may know a better path.
Her line that stuck with me: the way to look junior is to act like you already know everything. Curiosity builds the relationship. Pushiness teaches people to avoid you.
That is not manipulation. It is ordinary human interaction aimed at doing right by the customer and by the teammate.
AI accelerates whatever you already are
We got into AI's impact on software engineering, and Shaundai held two ideas at once.
On one side, agents can push more engineers toward an eagle-eye view: architecture, product intent, orchestrating work instead of only local ticket implementation. On the other side, she is already seeing the downside. If twelve agents chase patterns that are not really there, the repo fills with band-aids. Tribal architecture knowledge never made it into a prompt. Tokens get spent reconstructing chaos.
Her framing was helpful: AI does not invent a new failure mode so much as it accelerates the old one. A team lead with a group of interns already knows what happens when people copy a bad local pattern. Agents just close that loop in twenty minutes instead of a week.
So the durable work is still human. Software is for humans. Markets change. Taste matters. AI is a tool in the toolkit - a supplement, not a replacement. We need better practices for using it sustainably more than we need to wait for the model to get magically wiser about product judgment.
The homework
Shaundai's homework is delightfully concrete.
Try Josh Comeau's Whimsical Animations - she called out the CSS-based starter path - and do the first stretch without AI. Get outside your usual domain. Refresh the muscle of figuring something out yourself. Then notice how far you get before you decide where a tool would actually help.
That assignment is not anti-AI. It is pro-judgment.
