Challenge

If you work for one hour to build something that automates a one minute task, the 61st execution of that automation is pure time profit. If you can build it in seconds, then you have pure time (money?) profit from the first execution itself. Writing python scripts takes hours, our no-code tool reduces that to minutes, your ML models will bring this down to seconds. How will you build them?

You would be a good fit if you:

  1. Think deep learning is not the silver bullet to solve general machine intelligence, and have some opinions about what else might be
  2. Hate using the term "AI-based" to describe something because you know it's catch-all jargon for nothing in particular
  3. Have always wanted a Jarvis like personal robot to get your stuff done
  4. Have tried replicating in code at least one academic paper just to check if it works
  5. Can explain back-propagation to a 5 year old

What you'll be working on :

  1. Running experiments with different transformer architectures, adapting existing ones to our use case, cleaning and creating datasets
  2. Babysitting a large number of experiments at once, staring at plots and tweaking/re-launching what works
  3. Conceptualising new performance benchmarks & tests for the models - aligned business goals
  4. Interfacing with the product team to figure out how to integrate your work into Maya
  5. Occasionally writing articles/blog posts elaborating on your thought process

Some things we expect :