Neha Patki

Product, Applied ML, memes 🤪
Cofounder, maintainer of open source @sdv_dev
Formerly: PM @Google, AI researcher @MIT

[Synthetic Data Vault @sdv_dev]

Thanks to SDV Enterprise by @datacebo's SDV Enterprise, ING Belgium can generate synthetic SEPA payments on-demand, ensuring thorough testing and compliance w/o compromising data privacy.

Learn more from @n4atki, our VP of Product, and Wim Blommaert, Head of Test Data Management at ING Belgium:

https://t.co/StylFinHku

[DataCebo @datacebo]

Once you build a generative model to create #syntheticdata, you can sample “millions or even billions” of records of customers data, even if you train the model on just 10,000 customers' data, according to @kveeramac . “Enterprises can create a lot of data in their lower environments and not have to port the data from a large database, instead using this model that's maybe just a couple of gigabytes,” he says.

Check out @kveeramac full interview on the Cloudcast podcast w/Aaron Delp (@aarondelp ) & Brian Gracely (@bgracely), where he talks about how @datacebo is leading #syntheticdata efforts at: https://t.co/WZUgkOlNx8

#devops #bigdata #syntheticdata #generativeai #data #datascience #enterprisedata #tabulardata #predictiveAI #machinelearning #ML #generativemodels #MLmodels

[Synthetic Data Vault @sdv_dev]

We can't control the weather, but we ✨can✨ control our software: Instead of waiting for a storm, simulate a hypothetical one using #syntheticdata ⛈ ❄️

#SDV #datasimulation

https://t.co/6FBBn5foi3

[DataCebo @datacebo]

🎉2022 was an exciting year for us, our #opensource & enterprise SDV, and for #syntheticdata overall. 🥳⭐ Here’s a recap of our 2022 + some innovations to come in 2023👉 https://t.co/lg7MapTBqv #machinelearning #artificialintelligence

[Synthetic Data Vault @sdv_dev]

In true #syntheticdata fashion, we’ve created some ✨synthetic✨ Season’s Greetings.
Happy holidays from the #sdv team and looking forward to a great 2023!

[MIT - Data to AI Lab @lab_dai]

#SeasonsGreetings from @kveeramac and the team at the #MIT Data-to-AI Lab #AIforGood

[Synthetic Data Vault @sdv_dev]

Have you ever looked at a GAN & wondered what's happening inside? 👀
We took a peek -- and made it more interpretable 🙌
#syntheticdata #sdv #CTGAN

https://t.co/AFcbUMxq3r

[Neha Patki @n4atki]

A real world example of outliers! Here's a graph of my heart rate. Guess when I saw a spider? #datascience

[Neha Patki @n4atki]

When you finally get an iPad & get to subject everyone to your horrendous handwriting 😈
#datascience #illustrated

[Neha Patki @n4atki]

More thoughts in my latest article

👉 https://t.co/tRAI5YWEAh

#syntheticdata #sdv

[Neha Patki @n4atki]

I started out my career as an AI researcher. Now, I'm leading product at a startup, convincing people that you don't always need the fanciest ML.

Irony ... or just the natural result of a business-driven mindset?

[Stack Overflow @StackOverflow]

With AI and ML-powered coding tools on the rise, the data used to train these models are under scrutiny and raise major privacy concerns. We dig into how synthetic data presents a lower-risk solution that scales.

https://t.co/gmQ2rXAiyI

[DataCebo @datacebo]

4 ways to compare synthetic & real data with SDMetrics.

Explore the open-source library: https://t.co/mQrSakt2dR

#SyntheticData #DS #DataScience #BigData

[François Chollet @fchollet]

Having a solid product is good. What's better is to have a fast and reliable process for continuously improving it.

(This is why open-source has such an unfair advantage.)

[Neha Patki @n4atki]

Apple's iOS 16 "unsend" feature is a prisoner's dilemma:

1. If everyone upgrades, everyone gets to unsend 🙌
2. If I upgrade but you don't, you get to spy on my unsent texts 😡
3. If nobody upgrades, nobody gets to unsend 🤷🏽‍♀️

[Neha Patki @n4atki]

I'm not worried about AI taking over bc it just does what it's told. You need human creativity to caption that image: "me on a Monday, when my software isn't working"

[Neha Patki @n4atki]

The image below is AI-generated ⚠️
(I asked DALL E to create a displeased fish out of water, wearing a funny hat. did not disappoint) #dalle2 #OpenAI

[Neha Patki @n4atki]

My team, climbing out of the very special hell that is Python package mgmt, cutting 1 dependency at a time ✂️

#opensource #python #softwaredevelopment

[Neha Patki @n4atki]

@benhamner and also if you do it wrong, you may accidentally set something on fire 🔥

[Neha Patki @n4atki]

For ML products, meta data is such a great tool for debugging. The user doesn’t have to share actual (private) data, but still communicates the structure/schema

#syntheticdata #machinelearning

[Data Science Fact @DataSciFact]

One person's meta data is another person's data. It's all data, just different perspectives.