Adham Ghazali

Spent twenty years building hard machines. This is what I learned. And what I'm doing now. remoroo.com

@FarzaTV The healthiest founder mindset I’ve found is to race against yourself. Not competitors. Not other people’s ideas. Just find one area each week that can improve with real attention, improve it, and let that compound.

Mad respect for Julian, but I disagree on this one.

One thing I like about Founders Inc: they seem to have a real bias toward companies solving painful, physical-world problems. This is a great example.

They reinvented the hearing aid by studying the human ear Normal hearing aid: $4700 Theirs: $20

I used to think small startups should avoid depending on each other. AI changed that. Two well-aligned startups can now move like one company and go after markets that used to require an army.

@KaiXCreator Def robots doing stuff …

What will come after AI?

@hthieblot True. You just had unusually good taste.

@AdhamGhazali I thought you DMed me first lol

The next billion-dollar founder has 9 followers on X rn. I will find you & fund you!

This is what most robot demos look like before the demo.

Vision vs LiDAR, Tesla vs Waymo: youtu.be/V6tasPBGw6E?si…

Very simple

We built a dojo for research agents. Not “fix this bug” benchmarks. Robotics calibration. RL control. CIFAR speedruns. Genomics triage. HIGGS. Speech. Each task has a locked eval, one metric, and an append-only experiment log. Welcome to RooDojo: github.com/remoroo/Roodojo

@yousefhll will be happy to tell you all about it in person soon. coming to SF ...

@AdhamGhazali can't wait to see what's in store!

Spent twenty years building hard machines. This is what I learned. And what I'm doing now. remoroo.com

@omoshmoze This is cool. I will send you a dm.

@AdhamGhazali kindly visit 4dlab.xyz

I want to help robotics founders this week. Not with fundraising. Not with generic AI advice. Not with pitch feedback. With the painful part I know well: turning messy robot behavior into a faster experiment loop. Perception failing in edge cases. Motion planning stuck. Sim not matching reality. Hardware creating weird constraints. Metrics unclear. I’ve built robots since I was 12, scaled a self-driving company, and built one of the fastest strawberry-harvesting robots in the world. I can help you find the real bottleneck faster than another week of guessing. If you’re building a robot and stuck in the loop, reply with what’s broken or DM me.

@enguerandvitro Makes sense

@AdhamGhazali Exactly same source the grid enables TSDF temporal fusion + ray-cast free-space On glass rays often punch through to the things behind but multi-angle observations + probabilistic update converge on the actual surface Resolution is granularity fusion does the denoising

Every robotics startup talks about point clouds and neural nets We’re betting on 1 cm³ voxels instead It’s slower, older, and less flashy But on a glass facade with no texture, it’s the only representation that still works Sometimes the best answer isn’t new

@enguerandvitro So the real question is: why does the voxel representation remain stable if it’s derived from the same source: depth maps? Are you suggesting that the 1 cm³ voxel resolution itself helps suppress noise, or is there another factor at play?

@AdhamGhazali Thanks ! yes we've looked at Curobo The key is that on glass facades, texture is sparse and reflections are nasty, so a voxel representation stays stable where point clouds get messy

@yashhq_22 It does not matter because it’s a circle.

As a solo founder, what do you build first? - the product - the audience

@DanielSmidstrup It’s a circle man. Does not matter how you start!

As a solo founder, which one are you doing? - build first, validate later - validate first, build later

We’re hiring a founding engineer at @RemorooLabs. Backed by Founders Inc, working on one of the harder problems in agents: making LLMs run for real, not just answer in a chat box. remoroo.com/careers/foundi…

Coding agents are the first version because code is measurable. But can you let Claude Code, Cursor, or Codex run for a week? Not really. That’s the problem we’re solving at @RemorooLabs.

Remember when chatbots were a joke? Now every company has one. But I don’t think chat is the real interface. 🧵