gentube
Why we’re here
The internet got good at holding attention and bad at giving anything back. We think making things can be the most fun people have online. We want a billion people to feel that.
Cofounded by Mo, on the founding team at Maluuba (acquired by Microsoft), then Microsoft Research, and Josh, a founding engineer at Wish who helped scale it to 100M+ users and $2B in sales.
Open roles
Software Engineer (Generalist)
You build what people come here to play with, and the systems under it: 100,000+ images a day, edits that return in under two seconds, state that stays in sync on every player’s screen. You own entire features and the hard problems inside them.
What you’ll do
- Make generation fast: request pipelines, caching, GPU scheduling, model swaps.
- Build real-time systems: shared state, presence, play that updates live for everyone.
- Take a hard problem apart, find the smallest thing that answers it, ship it, measure it.
Technical requirements
- TypeScript end to end: React and Next.js on the front, a real-time backend like Convex behind it.
- Have built and run a production web app: auth, data model, caching, monitoring, on-call.
- Know how image and language models are served: latency, batching, cost per call. Python when the model code needs it.
- Can read a profiler, a slow query, and a funnel, and know which one to fix first.
We’re looking for people who
- Are strong general problem solvers. New domain, unfamiliar codebase, still get to the answer.
- Can take an idea to production in three hours.
- Have built something hard that real people used.
Machine Learning Engineer
You own the models behind every image. We run our own inference, generating 100,000+ images a day at a fraction of a cent each. Your job is to make them better, faster, and cheaper.
What you’ll do
- Own GPU inference end to end: serving, batching, autoscaling, cost per image.
- Work inside the models: fine-tune, distill, quantize, and cut latency without losing quality.
- Build the evals: automated checks for image quality, fidelity to each fandom, and safety, run against a baseline on every change.
- Get a new model, paper, or repo running in a day and decide whether it earns a place in production.
Technical requirements
- PyTorch and CUDA. Comfortable reading model code and kernels, and profiling where the time goes.
- Have served diffusion models or LLMs in production: inference servers, multi-GPU, containers.
- Hands-on with at least one of: LoRA or full fine-tuning, distillation, int8/fp8 quantization, attention or scheduler optimizations.
- Can reproduce a paper from the PDF and tell you what it got wrong.
We’re looking for people who
- Experiment with image and video models for fun.
- Have made something 10× faster or cheaper and can show how.
- Care how the pictures look as much as how fast they come back.
Content Creator
You own how gentube shows up on TikTok and Instagram.
We’re looking for people who
- Have grown an audience from zero and have the results to show it.
- Know what makes people stop, watch, and share.
- Move quickly when a format isn’t landing.
Tools: TikTok, Instagram, CapCut, PostHog, Discord.
What you get
- Top-of-market salary and full benefits.
- Meaningful equity in a company backed by top consumer investors.
- A place where play is the work.
To apply, email careers@gentube.app.