How I helped a hyper-growth AI startup grow from $2B+ into $10B+
A year shipping the table-stakes and enterprise features like IAM, Compute, Storage, Orchestration, turning a 3-month old nascent ML cloud provider product into feature-rich platform that companies like Rivian, Sony, Databricks, NVIDIA and Cartesia run their AI workloads on.
Company Overview
Crusoe is a vertically integrated AI infrastructure company that builds and operates data centers powered by sustainable energy sources. One of its verticals is Crusoe Cloud which provides set of developer surfaces to manage compute resources, including the Crusoe Cloud Console, CLI, APIs, Terraform, and SDKs. AI/ML engineers use Crusoe Cloud to run AI training, fine-tuning models, and running scaled inference. Crusoe recently ranked #3 as World’s Most Innovative Companies of 2026 after NVIDIA and Google.
Improved E2E Enterprise Onboarding Experience
Led the design vision for a unified enterprise onboarding experience, consolidating 7 fragmented workflows into a single cohesive journey for enterprise onboarding on Crusoe Cloud. This effort included multiple L-sized projects: Social Sign On, optimize email sign up, Auto-KYC, Stripe billing Integration and optimizing creation of first AI workload.
5 mins
TTT from 24+ hours originally
400%+
Increase in registered users/week
![Corporate Onboarding Flow [FUTURE].png](https://static.wixstatic.com/media/273081_d12576eb70894b418312b2c0abdaa15a~mv2.png/v1/fill/w_1412,h_285,al_c,q_90,usm_0.66_1.00_0.01,enc_avif,quality_auto/Corporate%20Onboarding%20Flow%20%5BFUTURE%5D.png)
IAM: Project Roles & resource-granular permissions
Took an undefined and complex IAM ask — RBAC vs. PBAC, project-level vs. resource-granular — and shaped it through PM whiteboarding, hyperscaler audits, and users-policy-roles mapping. Scoped a six-month vision back to a shippable MVP.
95.3%
MRR contributing customers use this feature
VM instance provisioning
Improved virtual machine instance creation process by tailoring to ML engineers needs and workflows, added feature capabilities, smart defaults, consumption models information and 7-page churns into a 1-click experience.
5 secs
TTT from originally 4.8 mins
100%
Reduction in support tickets related to unavailable resources
Driving AI-native design workflow
Leading the transition from Figma-centric pixels to repo-native design-in-code, reducing time to prototype and test ideas from weeks to hours.
