Robot Fleet Operations
Scaling autonomous robot deployment from a small pilot to 300+ units across 30 facilities — and reducing deployment time from 1 week to 2-3 days.
Robots in the real world are messy
Meta was deploying autonomous robots in warehouse and logistics facilities. The technology worked in controlled environments — but scaling to dozens of sites, each with different layouts, workflows, and constraints, was a different challenge entirely.
When I joined, deployment was a week-long process for each site. A team would travel on-site, manually map the facility, configure the robots, run tests, troubleshoot issues, and eventually hand off to operations. It didn't scale.
"Every facility was different — different floor plans, different workflows, different edge cases. We couldn't just copy-paste a deployment playbook."
Making deployment repeatable
I worked across the deployment pipeline — from robot mapping and configuration to deploy automation and operational analytics. The goal was to turn deployment from an artisanal process into a scalable system.
What I Worked On
- Robot mapping and localization workflows
- Deployment automation tooling
- Operational data analytics and dashboards
- Cross-functional coordination with hardware, software, and site teams
- Process documentation and training
Key Challenges
- Each facility had unique physical constraints
- Robot behavior depended on accurate maps
- Remote debugging with limited visibility
- Coordinating across time zones and teams
From 1 week to 2-3 days
By systematizing the deployment process — better tooling, clearer playbooks, automated validation — we reduced deployment time from a week to 2-3 days per site. The fleet scaled from a handful of pilot robots to 300+ units across 30 facilities.
What I learned
This role taught me that scaling isn't just about technology — it's about process, communication, and documentation. The robots worked. The challenge was making deployment predictable enough that we could do it 30 times without a specialized team on-site every time.
I also learned the value of operational data. When things went wrong (and they did), having good analytics meant we could diagnose remotely instead of sending someone on a plane. Visibility wasn't just nice-to-have — it was essential for scale.
Working at Meta's scale also taught me about cross-functional coordination. Hardware teams, software teams, site operations, logistics — everyone had different constraints and priorities. Making progress meant understanding those constraints and finding solutions that worked for everyone.