Join me at Dumme

TLDR: We're building an AI tool to extract interesting clips out of long videos and podcasts. I'm looking for smart people to join me.
[EDIT: I also have a co-founder now]
We want to help video creators move faster. Clips and shorts of long podcasts have a very high retention rate and usually account for more than 5x the views of the video they were taken from. A secondary channel like Lex Clips makes more views than Lex, the primary one, cumulatively.
Similarly, people consuming videos and podcasts often want to jump straight to the point to learn fast. Ever tried to learn about something only to find yourself lost in hours of off-topic content? We solve that. In addition to our clipping tool for creators, we're working on a mobile app that allows people to learn faster from videos, by having an algorithmic feed of semantically chained clips.
We focus on interesting educational content, not dancing videos. No cheap dopamine hits.
We were part of the Y Combinator W22 batch, I applied and got in with a different idea, but decided to pivot after several months to work on something that has AI at its core.
Dumme hasn't launched publicly yet, so we’re still very early. Joining us now will make you one of our first hires, which also means that you’ll help us:
- Set a direction for the products (it's always likely to morph at this stage!)
- Build and scale our infrastructure (AI inference, rendering and streaming are expensive, we need optimization at scale!)
- Make future hiring decisions (compound effects are great!)
- Define our company culture (Still a blank canvas)
We're mostly looking for technical folks to join us, but are open to different profiles.
Some things you’ll do with us if you're technical:
- Work on our indexing pipeline. We take long videos and run them through a bunch of models to transcribe them, visually analyze them, identify the important moments, generate clips, etc.
- Improve our rendering engine. We are constrained by FFmpeg and need to get GPU-accelerated rendering to work inside Docker containers to be able to edit clips fast. This is essential if we want simple things like portrait mode, subtitles, effects, etc.
- Do some AI mad science and try to implement bleeding edge research for practical use-cases. We're currently playing with things like LaViLa and Temporal Alignment Networks.
- Build a reliable and scalable video streaming infrastructure, similar to Mux, but cheaper. We currently have something minimal but it works. It's a bit unclear whether it can scale, but we don't care for now!
- Help us with our mobile app. React Native is a pain if you're not used to it. We're trying to ship a Testflight. We thought using React Native would be faster because we eventually want an Android app. But of course, we're open to experimenting with other tech stacks for mobile.
- Build a lot of things people want!
Some things you'll do with us if you're not technical:
- Help us onboard video creators and podcasters. Figure out what they need.
- Define our marketing and growth strategies as the product evolves.
- Do a lot of schlep with us.
- Make sure we do the "business" stuff correctly!
The interesting thing about joining Dumme is that our core tech sits at the intersection of AI and video and requires us to continuously solve a lot of challenging problems to make everything work properly. You’ll do a bit of everything for now, but as the team grows, you’ll be able to focus on whatever you like the most (Most of the time this is also what you’re good at!).
You can email me at [my first name]@dumme.com or DM me on Discord (merwane#8721) and I'll get back to you. You don't need to send a resume or anything, but I’d love to see some things you’ve shipped before. It can be anything: OSS, side projects, a product at another company, etc. If you're not technical, just tell me what you've been up to so far. Bonus point if you're in London (where I am) or the SF Bay Area (where I'm hoping to move!). We're open to working remotely as long as you’re in the United States, Canada or Western Europe due to timezone convenience (and regulatory stuff).
FAQ
Have you raised any money?
Yes! Happy to share more details over a call.
How will Dumme make money?
We're planning on charging creators a monthly subscription for using Dumme, in addition to showing contextual sponsored content on the consumer app.
How can I join Dumme?
We’ll jump on a call and discuss various topics (including AI!) to make sure we’re on the same page. If we both like each other, you'll join me full-time for a couple of weeks (2-3 weeks tops; fairly compensated), to make sure that we’ll be able to work together for the years to come. If this isn’t an option for you, we can simply keep talking until we're both convinced you should join Dumme.
What’s your tech stack?
We mainly write code in Python, and use Docker to orchestrate our components, which are structured as a distributed monolith. This means that we can move faster than with something like a microservice architecture, while keeping some components work independently. Each component can be written in a different language and can have an independent runtime. Our indexing pipeline consists of running long videos through multiple AI models. No worries if you don’t know a lot about AI, we’ll help you ramp up! We use a fair amount of big libraries and frameworks, including PyTorch for AI stuff, FFmpeg for all things video, etc.
We also use Nuxt and React for our frontends, and our mobile app is being built using React Native.
Does the tech work? Is it live?
Yep, the tech works, but the product isn't live yet. We can give you a demo over a call! The interesting thing about the underlying technology we create and use at Dumme is that it is universally useful and isn't limited to our use case. If we can make a computer understand something just by looking at it, we can make it understand the world. We want to create a future where machines understand us.
What’s the work/life balance like at Dumme?
We’re still very small, but we don’t want to hide the fact that we spend more time working than average. More hours make an enormous difference at this stage. I personally often find myself working at unconventional hours. There are no rules when it comes to this. We’d rather have you work with us when you’re effective, not when we think you are. We have a small office in London and we'd rather work with you in person as we find it hard to be productive and spontaneous in a remote environment.
Finally, we’re a mission-focused company. We want to use AI to help people create, learn and explore faster. As a consequence, we don’t engage in politics or advocate for anything that isn’t related to our core mission. We’ll hire you regardless of your beliefs, political orientation, and of course, everything that you were shipped with.
Note: It's pronounced dummy!