Machine Learning made simple.


are hidden in data.

In this era of data accumulation, Kashf learns to find answers in data from users without any expertise.

Target Illustration


Bootstrap Learning

Kashf learns to recognize concepts using a few exmaple texts, images or vidoes and finds new instances of that concept. Then, user corrects these predictions and Kashf learns from its mistakes. In this process, Learning never stops.

Labeling Hard Examples

There are instances of a concept which Kashf confiedntly recongizes, but some examples are more troublesome. Kashf asks label of these important instances from user. So, Kashf does more with less user labels.

In Detail Evaluation

Evaluation comes before every improvement. Kashf starts learning process with model selection and parameter tuning. After training, Kashf reports overall and detail evaluation of trained model.

Serving At Scale

Through learning process, Kashf is always ready to find concept instances at scale. It uses distributed processing to harness modern hardware's computation power of cloud solutions.

Example Concepts

Kashf learns to find anything:

Trump is not only thin skinned and petty, he poses an increasing threat to the free press. #HeroJimAcosta

@RyanAFournier Good he crossed the line. President Trump deserves respect ✊. Ban CNN

@teslamodel3fan @elonmusk @Tesla Or at a minimum @tesla can put a button under the backup camera display to fold mirrors when in reverse.

i can’t wait to own a @Tesla when i’m extremely finically stable

Get in touch

Send a message,
right away

kashf [at]

+98 21 9100 33 03

No. 9, Hosseinali St., Farjam Ave., Narmak.
Tehran, Iran


Send a message,
right away