Krishna Gade Profile picture
Jan 20 16 tweets 7 min read
I was a first-time founder when we started @fiddlerlabs, and we founded it with a big mission to build trust in AI.

This is a thread on the journey we went through in the last 3 years running the company and the lessons we learned. 🧵
First up, the startup journey is not for the faint-hearted. The highs are super high and lows are super low. I am lucky to have got a great co-founder in @amitpaka who has been there throughout this journey.

So, if you’re thinking of getting a co-founder - you should! /2
When I was starting @fiddlerlabs, the first person I met was @amitprakash whom I consider a mentor of sorts.

He told me - "if you’re planning to do a startup you should take some risks, meaning go after a big hairy goal instead of doing something incremental". /3
It turns out helping enterprises build trustworthy AI is a big hairy audacious goal, that I don’t know if we will ever accomplish fully.

One thing about BHAGs is that they inspire you and motivate you especially when you’re down in the dumps.

So go after a BHAG! /4
We got really lucky to get a great set of investors in @lightspeed, @lux_capital, and @insight, It all started with our chance meeting with @semil.

Getting @semil was game-changing for us because not only does he back his companies, he actually hustles for them #Respect /5
A great team is critical for any startup to be successful. When you have a long-term mission - it is critical to hire missionaries and weed out mercenaries who are there just for title/pay, etc.

It is a continual process, you have to keep improving the team as you go. /6
With @fiddlerlabs we are defining a category in ResponsibleAI and MLOps.

Creating a new category is hard! It requires a combination of non-obvious product insight, unique GTM, and an ability to weather skepticism.

You've to keep evangelizing and iterating with patience. /7
Early on, we studied regulations for Model Risk Management in Banking that mandate monitoring, explainability of AI.

We used to be the only MLOps startup that attended MRM conferences and this helped our product and GTM a lot.

Now, Fintech is a major market segment for us. /8
The overall market turned a corner 18 months ago, and MLOps got on fire as various other verticals started looking for ML Monitoring and Explainability tools.

It is important to start with a market insight but also have a flexible playbook when things change in the market. /9
Product-led growth is the new thing in the enterprise space. Gone are the days when your sales team can sell software that users hate to use. Modern enterprises shop around with a consumer mindset.

UI/UX polish, ease of use, onboarding experience are supercritical. /10
However, getting the product experience to the T in a new category can be difficult to get right from day-0.

Because no one before you has built a product in this category. There is no playbook here, you have to figure it out as you go. /11
We learned that it is critical to have a set of early adopters, who are trying the product and giving feedback. Otherwise, it becomes hard for your team to prioritize new features, bug fixes, and scalability issues.

So, make sure you establish a customer council early on. /12
Finally, it is super important to get your company culture right. IMO, culture is not what you preach, it is not what you say at your all-hands, it is not the list of values on your website.

Culture is what your company does on a day-to-day basis. /13
It is how the people in your company communicate and work with one another, celebrate successes, and handle failures.

Make sure you hire, retain and promote culture carriers. Because the teams that eventually win are the ones that have created a strong resilient culture. /end
Thanking all our investors and correcting a few twitter handles. @lightspeedvp, @BloombergBeta, @insightpartners, @Lux_Capital, @haystackvc, @zgulamali, @LockheedMartin Ventures
Correcting the Twitter handle for Amit Prakash @amitp42.

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Krishna Gade

Krishna Gade Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @krishnagade

Dec 2, 2021
Zillow has recently shut down its AI-enabled iBuyer program that overpaid thousands of houses in summer 2021, along with laying off 25% of its staff.

Are teams managing these risks w/ AI well?

THREAD: How to build a robust Model Risk Management (MRM) process in your company?
1/ Zillow invested a ton of money into its AI-enabled home-flipping program called Zillow Offers.

They bought up thousands of houses per month, whereupon the homes would be renovated and sold for a profit.

Unfortunately, things didn’t go to plan.

bloomberg.com/news/articles/…
2/ Zillow's CEO recently said that they failed to predict house price appreciation accurately:

“We’ve determined the unpredictability in forecasting home prices far exceeds what we anticipated.”
wsj.com/articles/zillo…
Read 15 tweets
Oct 7, 2021
I was an engineer on Facebook's News Feed and this is NOT how recommender systems work.

While users can set some explicit preferences, implicit user activity on the app is the bulk of the signal that gets fed into the AI systems which control & rank the feed. /thread
So, if you're engaging with a certain type of content of a certain set of friends - the stories from those sources get ranked higher than others. This is true with Facebook or YouTube or any other recommender system. /2
These systems take activity events from a user’s activity history as input and retrieve hundreds of potential candidate stories/videos to show to the users in their feeds. /3
Read 7 tweets
Feb 11, 2021
I was an eng leader on Facebook’s NewsFeed and my team was responsible for the feed ranking platform.

Every few days an engineer would get paged that a metric e.g., “likes” or “comments” is down.

It usually translated to a Machine Learning model performance issue. /thread
2/ The typical workflow to diagnose the alert by the engineer was to first check our internal monitoring system Unidash to see if the alert was indeed true and then dive into Scuba to diagnose it further.
3/ Scuba is a real-time analytics system that would store all the prediction logs and makes them available for slicing and dicing. It only supported filter and group by queries and was very fast.

research.fb.com/wp-content/upl…
Read 11 tweets
Nov 25, 2019
With the last week's launch of Google Cloud’s Explainable AI, the conversation around #ExplainableAI has accelerated.

But it begs the questions - Should Google be explaining their own AI algorithms? Who should be doing the explaining? /thread
2/ What do businesses need in order to trust the predictions?

a) They need explanations so they understand what’s going on behind the scenes.

b) They need to know for a fact that these explanations are accurate and trustworthy and come from a reliable source.
3/ Shouldn't there be a separation between church and state?

If Google is building models and is also explaining it for customers -- without third party involvement -- would it align with the incentives for customers to completely trust their AI models?
Read 11 tweets
Oct 10, 2019
We've been working on #ExplainableAI at @fiddlerlabs for a year now, here is a thread on some of the lessons we learned over this time.
2/ There is no consensus on what "Explainability" means. And people use all of these words to mean it.
3/ However, one thing is clear:

"AI is a Black-Box and people want to look inside".

The reasons to look inside vary from Model Producers (Data Scientists) to Model Consumers (Business teams, Model validators, Regulators, etc).
Read 14 tweets
Sep 19, 2019
It is amazing to see so many applications of game theory in modern software applications such as search ranking, internet ad auctions, recommendations, etc. An emerging application is in applying Shapley values to explain complex AI models. #ExplainableAI
Shapley value was named after its inventor Lloyd S. Shapley. It was devised as a method to distribute the value of a cooperative game among the players of the game proportional to their contribution to the game's outcome.
Suppose, 10 people came together to start a company that produces some revenue. How would you distribute the revenue of the company among the 10 people as a payoff so that the payoffs are fair and appropriate to their contributions?
Read 11 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us on Twitter!

:(