Advice from the Top 1% of SpaceX Engineer | Sift, Karthik Gollapudi | EO

SpaceX's urgency, packaged as software for hardware teams

Karthik Gollapudi spent four and a half years on Dragon flight software at SpaceX, and what he took away was a standard rather than a story. SpaceX ran on urgency and a high iteration rate. First principles questions there stop only at physics, and he argues that way of working is learnable if you see it up close. Sift sells it to hardware teams as data tools for safer machines, starting embedded inside its first customer's office.

Takeaways

• See greatness up close before founding. He advises against starting a company immediately. Work with the smartest people you can find and spend multiple years on one hard problem rather than a year each at several companies.
• Prestige is a poor proxy for learning. The people who made a place successful may have moved on, and it may be larger and more process-heavy. He chose SpaceX over Stripe and Palantir because it pitched mission rather than benefits or work-life balance.
• SpaceX ran on urgency and fast iteration. The biggest thing he learned at SpaceX was what good looks like. What stood out was urgency and a high iteration rate, and the results were visible in the largest rocket ever launched and landed.
• Ask why until physics is the only blocker. First principles thinking is the habit he rates highest. Wernher von Braun had every engineering team at NASA write weekly notes because risk assessments were downplayed as they passed up the management chain.
• Sift sells SpaceX discipline as software. Sift builds data tools for modern machines, productising what Gollapudi learned about building safer and more reliable hardware. It raised about $25m, with a Series A led by GV and an earlier round led by Quikka Ventures and Riot Ventures, and had seven announced customers.
• Sift was built inside its first customer. Parallel Systems, the autonomous rail company, was the first customer. Sift rented a room in its office and its early roadmap was driven entirely by that feedback. Later customers came through the SpaceX alumni community and investor introductions.
• Everything comes down to people. That is Gollapudi's biggest lesson from building Sift. He warns against high-profile advisers who are not vested in your success, and embedding engineers with customers came from Elon Musk wanting the engineer on the factory floor.

The Dragon parachute anomaly

1. One of four parachutes on a Dragon mission opened late, at low altitude.
2. NASA grounded the following flight, scheduled a week later, until the issue was understood.
3. The capsule landed in the Gulf of Mexico with no internet connection, so SpaceX asked the Starlink team, then still experimental, to switch on cells over the Gulf.