Experimentations are the bread and butter of most great product companies (albeit google takes it up a notch-where you could call it insanity). Experiments need to be designed with rigor so that there is confidence that the changes are right for the customers and the business as well.While Randomized controlled experiments are the gold standard for establishing causality, but sometimes running such an experiment is not possible. Such situations can arise:

  1. When the causal impact to be tested is not in companies control.
  2. When establishing a Control may incur too large an opportunity cost since they do not receive the…

Coming up with a great set of metrics is an art in itself — they can lead to better Product Decision, makes it easier to align all stakeholder, and discovery around how your product is being used.

Each Product will have 3–5 Key Metrics, that can provide a holistic view of the product’s value to the end user. (Read more here). The key metrics should act as proxies to the business value the product is generating, and also be responsive to product changes. A good product company prioritizes features that optimize its key metrics. …

checkout https://cutouts.app/

Two weeks earlier, we launched Cutouts on producthunt ( We were in product hunt’s top 5 product of the day! Check us out here.) We got a lot of requests from people to write about building a ML model for browser- so here we are!

We wanted designed cutouts to be privacy-first, but also give users an amazing user experience. Tagging and Searching had to be taken care using Machine Learning, and the given all the constraint, it was clear to us we had to run the ML models on the browser. In comes, Tensorflow.js, which let’s you use ML…

Photo by Scott Webb on Unsplash

My friend works at a company where they were still using email for their day to day communication, and where suffering from huge communication gaps. (Emails getting missed, important messages getting lost in the thread, no transparency). I suggested them to try out the free version of Slack, and they are instantly hooked. They will soon be moving to a paid version as now all their communcation channels — the off topic groups on whatsapp, customer service messages from intercom, clients, are now connected to Slack. …

Photo by Jared Rice on Unsplash

There have been so many times in our lives where we’ve felt very alone and miserable. But very slowly we have learnt how to take a few things in our control. And that is the beauty of being a human. We can come out as a better version of ourselves even after going through tough times. This is a curation of what I have learnt and what people around me feel is important.
(I am also volunteering as a listener at a self help community, if you are going through a tough time, we are there for you! …

1) The innovation

Empower Data Analyst/Software Engineer/Data Scientists to build state of the art computer vision solutions

2) Why do it? (The real pain)

Current deep learning methods require a huge amount of data and compute. Most Companies can benefit from having some kind of in-house CV related algorithms. Yet the expertise and money required to build a production-grade solution is too high for most companies. Even for an experienced Data Scientist, a quick prototype (considering data is available) can take 3–4 days. Most people also do not care about the algorithm used, they only care about the end results. …

The top metric at Kubric for the Data Science team is — “How do we reduce the amount of time spent by a designer on doing grunt work?”. We want to eliminate the grunt work that goes into micro-editing and quality checks so that designers can work on the stuff that truly matters to them — designing good content. Around this, we have created a set of micro-tools to help designers really get rid of that grunt work.And this is today powering thousands of creatives across different customer verticals from food to fashion and more. …

Tensorflow object detection API is a powerful tool for creating custom object detection/Segmentation mask model and deploying it, without getting too much into the model-building part. TF has an extensive list of models (check out model zoo) which can be used for transfer learning. One of the best parts about using TF API is that the pipeline is extremely optimized, i.e, your resource is not underutilized. If you have ever used Keras generator and trained a model, you will find how underutilized your CPU and GPU some times are, as data reading might become a bottleneck to training.TF …

Variational Dropout was introduced by Yarin Gal as a mean to estimate uncertainties in prediction by neural networks. While this blog isn’t about variational dropout(Follow Yarin Gal’s Blog post for that), it is about the results that I saw while playing around with Variational dropout.

If you are aware of how Variational Dropout works, it requires the dropout to be on during the testing phase also. There are two ways to do this in keras.

def PermaDropout(rate):
return Lambda(lambda x: K.dropout(x, level=rate))


Dropout(0.5)(x, training=True)

Using this, I quickly trained a 3-layer CNN network on MNIST.

inp = Input(shape=input_shape)
x =…

Here are some examples I generated to see Adversarial Examples of Vital Signs Time Series. You can see how the Time Series changes. It is no more the same example, it can be very well considered a different and independent time series vital sign sample. Unlike in images, where a human can still tell the target label easily.

Adversarial Images can be correctly classified by humans

Adversarial Examples were created using Jacobian Saliency map Attack with theta value as 0.4 and gamma as 0.2.

Aashay Sachdeva

Connecting the Dots.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store