Karthik Ravi headshot

Karthik Ravi

I'm building CueCloud. We help teams run AI on their own hardware.

How I got here

I used to work on ML inference at Meta, tuning transformers for the ranking system. I spent a lot of time figuring out how to make models run faster on the hardware we had. These days, I'm back at it with CueCloud.

In between, I bootstrapped two companies, built Stitchology with my co-founder, and went through an acquisition. I also led engineering at QuestionPro and Skillfully. Along the way, I had to learn how to sell what I was building, too.

We started CueGrowth to build AI tools for sales teams. Once we were running agents ourselves, I kept getting pulled into the cost of running them and the hardware underneath. That became CueCloud. We made the pivot in May 2026.

Why I'm working on this

There's a lot a team could do with AI if it could use their code, internal tools, and research. But handing all of that to an outside service isn't always an option.

I think a team should be able to run this stuff in-house without building a data center or hiring a whole team to keep it running. We're working on making that easier.

What we're building

CueCloud

CueCloud helps teams run AI where they work, on hardware they own. We use compact Apple silicon systems, get the models running, and connect them to the code and tools the team already uses.

Vault

The hardware and software to run models in-house. We help you get it set up and keep it running.

Blitz

Where your team works with coding agents, CueCode IDE, and your own tools, all inside your environment.

We're hands-on with the setup: picking hardware, tuning models, connecting tools, and trying it all on real work. Getting a model to run is the starting point. I care about whether someone actually wants to use it the next morning.

Visit CueCloud →

Where I've worked

  1. May 2026 – Present

    CueCloud

    CEO & Co-Founder

    What I'm working on now: helping teams run AI on their own hardware. This grew out of CueGrowth, and we made the pivot in May 2026.

    Visit CueCloud

  2. Jan 2025 – Apr 2026

    CueGrowth

    CEO & Co-Founder

    We built tools to help sales teams research buyers and run outreach. Running AI agents ourselves got us thinking about the infrastructure, which eventually became CueCloud.

    Visit CueGrowth

  3. Jun 2023 – Jan 2025

    Skillfully

    CTO

    I led the technical side of Skillfully. We helped employers see how people handle real work through job simulations.

    Visit Skillfully

  4. Jan 2023 – Jun 2023

    QuestionPro

    VP of Engineering

    After the Stitchology acquisition, I led engineering at QuestionPro, which builds survey and research software.

    Visit QuestionPro

  5. Dec 2021 – Dec 2022

    Stitchology

    CTO & Co-Founder

    My co-founder and I built Stitchology to turn what people were doing on a site into predictions. We combined session replay, analytics, and ML to spot patterns in user behavior and predict discounting patterns. I built the technology as CTO, and we took the company through an acquisition.

    Read the acquisition announcement

  6. Aug 2018 – Nov 2021

    Meta

    ML Inference Engineer

    I worked on ML inference, tuning transformers for the ranking system and figuring out how to run them more efficiently.

    Visit Meta

Education

  1. 2019 – 2020

    Stanford University

    M.S. in Statistics (Drop out)

    Started the master's program and later dropped out.

  2. 2015 – 2018

    University of Illinois at Urbana-Champaign

    B.S. Computer Science + Statistics

    Studied computer science and statistics.

Thoughts

I write The AI Company Builder Memo. Notes on building AI products, finding customers, and figuring things out as I go.

Read on Substack →