Approaching Four Years: Reflections on the Meta Journey
Three and a half years building VPS at Meta - from inception to platform. Considering what's next.
Somewhere in year three I noticed I had gotten better at my job while doing roughly the same job. July 2024 will mark four years at Meta, VPS is now a platform serving millions, and that observation is why this reflection needs to be an honest one.
What We Built
VPS at the end of 2023: integrated in Quest 3 and Quest Pro, 5,000+ mapped locations globally, millions of daily queries, 100+ active developer partners, and a team of 75 running it all. Concept to critical infrastructure in four years.
Career Evolution
I joined in 2020 as an individual contributor proving the concept. By 2021 I was a first-line manager shipping the MVP, by 2022 a senior manager scaling both the team and the system, and 2023 brought director-level scope, strategy, and partnerships. The role changed more than the technology did.
What I'm Proud Of
The system works: real users rely on VPS daily for real experiences. The team I built has strong leaders, a strong culture, and a sustainable pace. The privacy-first architecture proved you can build spatial AI without surveillance, which is the case I made in 2021 before it had to survive two consumer launches. And the synthetic data work extended what I learned at Magic Leap to Meta scale.
What remains unfinished?
Indoor, first: still not cracked, still the prize. Three Decembers running I have written indoor into the next year, last year's included, and the 2021 plan for closing the multi-floor and similar-room gaps has not closed them. This post does not put a year on it. Universal coverage, second - most places remain unmapped, and the long tail is very long. True real-time, third: latency is good but it is not invisible yet. And the economics: mapping costs are still too high for long-term sustainability.
The Learning Curve
Honest assessment: my learning rate has slowed. Years one and two were steep - new domain, new company, new scale. Years three and four have been incremental: optimization, polish, politics. I am better at what I do, but I am doing similar things.
Thinking About What's Next
The questions I keep circling: Is there more I can uniquely contribute here? What would I regret not trying? Where could my experience have maximum impact? I have not decided anything, but I am thinking seriously.
The AI Moment
The timing matters. We are at an inflection point for AI - LLMs are changing everything, multi-agent systems are emerging, AI-native products are suddenly possible. My perception and spatial expertise is valuable, but the frontier is moving, and I want to be at the frontier rather than optimizing behind it.
Gratitude
Whatever comes next, Meta gave me the resources to build at scale, the team taught me leadership, and the problems stretched me technically. Four years well spent.