Working Backwards & Forwards: 5 Years, 2 Cycles & 1 IPO at Opendoor
After 5 incredible years at Opendoor, today is my last day. Having shared my decision with the rest of the leadership team a few months ago, I wanted to take a moment to reflect on the journey. As any good applied research leader would, let me (over-)simplify for clarity:
2020-21 (AKA "Wait, we're doing WHAT?"): Started as an advisor making color-coded trade-off tables from the sidelines. Joined full-time because, well, the tables weren't enough. Made 17 contingency plans. Threw them all out. Built new ones. Convinced ourselves we weren't crazy. IPO'd. Quintupled revenue ($2.6B ⇒ $8.0B ⇒ $15.6B!). 🚀
2022 (AKA “So, about those interest rates…”): Enter the fastest series of rate hikes in 40 years. Installed more than a few "days since last incident" boards. Worked backwards from market fundamentals and forwards from algorithmic frontiers. Learned there's only so much you can do when "once in 40 years" happens twice in 2 years. 🤦🏾♀️
2023-2024 (AKA "Hold my whiteboard marker"): Merged engineering, product & AI/ML research into one beautifully chaotic family. Shipped breakthrough AI/ML innovations at a pace that'd make even the most cynical Twitter/X ML threads proud. Proved that anti-fragile isn't just a fancy word in library books (≤1%/yr regretted attrition!). Turns out great teams really do get stronger under pressure. Helped re-start growth. 🚀
Working backwards from what makes great R&D organizations tick: you need technical excellence that delivers real impact, teams that grow stronger under pressure, and the ability to unite different perspectives toward an ambitious goal. And sure, there's some secret sauce as well: maximizing throughput beats minimizing latency, designing uncorrelated risks in research portfolios is how you beat the house, dancing the three-step between commercial viability ⇔ technical feasibility ⇔ org health is art not science, etc. But mostly, you need extraordinary people who somehow make the impossible look routine. I've been incredibly fortunate to work with some of the brightest minds in the AI/ML industry who showed up every day ready to tackle the seemingly impossible challenges I threw their way.
To the incredible team who turned my grossly reductionist frameworks into reality (and politely nodded at all my "Kushal"-isms) — thank you. You taught me more than any color-coded table could capture. The resilience, creativity, and sheer determination you've shown through every market cycle is inspiring. I'm excited to see what the team will achieve next — the foundation we've built together is just the beginning.
What's next? Well, me being me, let’s work it backwards and forwards:
Step 3: Tackle another "impossible" problem
Step 2: After I figure out which one
Step 1: But first, a (doctor-mandated) nap...
So, funny story: I broke my ankle last week. Between surgery and the cast, I'm basically under doctor's orders to relax for the next 3 months.
Kushal, this is a beautiful reflection — warm, lucid, and unmistakably you. The way you translate contingency plans and color-coded tables into lived resilience is exactly what makes you such a rare leader. Most people talk about antifragility as a buzzword; you’ve clearly lived it with your teams through some brutal cycles.
The dance you describe — commercial viability ⇔ technical feasibility ⇔ org health -- is not just a startup truth, it’s a civilizational one. If more institutions could hold that balance, we’d all be in a stronger place. Wishing you deep rest, quick healing, and plenty of time with family (and the puppy).