@shub777 Profile picture

shubh

@shub777

Formerly at @AmericanExpress @JPMorganChase | AI Scientist & Engineer | Global Citizen now in Australia| Proud Father and Husband 🌏🤖

Joined January 2010
Similar User
Keaton Patti photo

@KeatonPatti

Nabla photo

@nabla_ai

Frank Willett photo

@WillettNeuro

AllenNLP photo

@ai2_allennlp

François Fleuret photo

@francoisfleuret

Abubakar Abid photo

@abidlabs

Phani Srikanth photo

@phanisrikanth33

Thomas Kipf photo

@tkipf

Rasa photo

@Rasa_HQ

Justin Johnson photo

@jcjohnss

Sara Robinson photo

@SRobTweets

Deep_In_Depth photo

@Deep_In_Depth

Michael Galkin photo

@michael_galkin

SkalskiP photo

@skalskip92

Martin Görner photo

@martin_gorner

I hardly click on any of the email newsletters that come in unless its by @asmartbear . Other gem of a read longform.asmartbear.com/bad-advice/ .


I do reckon e-commerce businesses have a content discovery problem which would be resolved in a big way as the UX moves more towards conversations. Here we have an easy prototype for FermiAI which we have been building to showcase the possibilities from shop browsing to payments


As funny as it is i have a huge respect for everyone in sales/ distribution particularly b2b sales. you don't realise how hard the work is if you don;t work closely with them.

Tweet Image 1

This is a great post by Jason Cohen of when to accept and when to ignore Wisdom of the crowd. Most often than not great ideas will be which would be loved by 5% of population but hated by 95%,. buff.ly/3ziSGdE


A fun app I am building to learn how good I and my friends and family are at identifying when something is AI or human generated. Anyone else interested in giving this a shot?


The best SAAS help business owners automate workflows, specially those which in the past might have needed dozen developers and a few deployment engineers. The KPI that is always front and center for me while building any software like FermiAI is, can everyone deploy it?


When solving problems, choosing the right evaluation metric is key! Consider: Base event rate (imbalanced data?) Cost of errors (false positive vs false negative) Rank-order vs calibrated results Focus on Top K or all predictions Problem stability (future distribution shifts?)

Tweet Image 1

In one of the past post we were working on metrics we could optimise to optimise the b2b sales funnel. This one is about which Machine learning models could help you with the optimisation. This will have a huge variance based on business stage. Any techniques I have missed?

Tweet Image 1

Here are 16 key metrics that I’ve found to be game-changers across the B2B sales funnel. These are critical for any company distributing products in the B2B space. Any misses?

Tweet Image 1

An age old problem which I still observe heaps of data folks concerned about is which metric should we use ? choice of metric more than anything is determined by costs of catastrophic failures, tfailsafe rules we have covering the ML models and what risks are we ready to take.

Tweet Image 1

One of the hardest skills to master for every data scientists is how the raw variables(specifically categorical variables) should be transformed for usage into model development for better predictive accuracy and long term maintainance

Tweet Image 1

Working with numerous B2B companies, optimizing sales pipelines, I've learned a powerful truth: it's not just about creating the best optimizations, but about knowing the right metrics to track and adjust. 📊. Below are 16 which I have found to be the most optimum for b2b distro.

Tweet Image 1

While everyone on X is going nuts with o1 and what @OpenAI is going to release next. What do folks who are building with LLMs reckon about this COT experience ?


Loading...

Something went wrong.


Something went wrong.