"What you all have done in a few days, we haven’t seen in 40 years!"

If you write Python for data, ML, generative art, or heavy compute, you know the bottlenecks: your local machine slows down, GPUs cost money, and environments get messy. Now, with Google’s newly launched Colab extension for VS Code, you can run notebooks with cloud GPUs and TPUs directly inside your favorite editor, all while keeping your workflow modular, version-controlled, and lightning fast.

The landscape of conversational AI is rapidly evolving. Conversational AI is no longer just about simple Q&A. Modern agents need to understand dialogue history, integrate external tools, adhere to safety protocols, and ideally, communicate across language barriers. This post provides a technical deep dive into "Critic's Cut", a movie review agent built to embody these principles using:

Say goodbye to overpacking and hello to perfectly curated travel wardrobes! This project leverages the power of AI to create personalized packing lists that consider your destination's weather, your personal style, and planned activities, ensuring you're prepared for anything the trip throws your way.

Ever wondered which Large Language Model (LLM) stands out not just for providing accurate responses, but also for being clear, respectful, and detail-oriented? With the ever-expanding list of LLMs, each boasting different configurations and capabilities, it is crucial to have a way to quantitatively assess their performance. The goal is to help us select the best model for building applications that prioritize precision, safety, and ethical considerations.

In the realm of AI-driven interactive storytelling, the fusion of Retrieval Augmented Generation (RAG) and Google's Gemma language model opens new avenues for creating engaging narratives and educational experiences. RAG, often likened to an "open-book" approach for AI, empowers models like Gemma to access and synthesize vast amounts of information to answer specific queries or weave intricate tales.

Are you tired of staring at a blank screen, struggling to craft captivating headlines and categorize news articles effectively? Say goodbye to news copywriter's block as we harness the power of Gemini Pro. In this article, we'll explore how this cutting-edge technology can supercharge your productivity and creativity, helping you craft compelling headlines and easily categorize news articles.

I recently came across a research publication titled The Shift from Models to Compound AI Systems on the Berkeley Artificial Intelligence Research (BAIR) Lab's website. Being on a time crunch I wanted a quick gist of the blog post without getting into the nitty-gritties. To make my work easier, I leveraged the capabilities of Google's Gemini 1.5 Pro to help me in this process of summarization. Once I got a rudimentary idea of the topic, I dived deep into it, and undoubtedly BAIR has done a great job with this 👏🏻

In the vast and ever-evolving landscape of online advertising, campaigns come in various forms, each with its unique approach to capturing the attention and imagination of internet users. Here is a fascinating spectrum of ad campaigns, from general to personalized and thereafter hyper-personalized. 

Last year, I co-authored a book chapter titled Do Not ‘Fake It Till You Make It’! for Springer Nature Group's Deep Learning for Social Media Data Analytics book series 📕 The publication walks through a comparative study of Deep Learning models to approach the tasks of identifying phony information, verifying the validity of various claims and facts, catching fake content, and so on. It eloquently analyzes the definition of Fake News specifically clickbait, hoax, satire, propaganda, hyperpartisan, and deepfakes in the world of social media, the various forms it can take, what causes its spread, and what are the rudimentary signs of such fake news. It further discusses the limitations of the detection algorithms with insights into the fairness, interpretability, and accountability along with providing pointers for the readers regarding emerging trends in this domain.

How do you keep up with the countless innovations and state-of-the-art R&D happening? This is one of the most commonly asked questions. 🎙💭🤔

Did you know that the ML benchmark datasets we heavily rely on like MNIST, ImageNet, CIFAR, etc., can have thousands of errors in their labels? 😱 Check out labelerrors.com for a few examples.

I just read this very interesting article -- it talks about the life of human (especially women) data annotators based in small towns and villages in India. They are the lifeline behind the most unglamorous part of the AI pipeline: data annotation, or labeling. I also learnt that India is one of the world’s largest markets for data annotation labor!

Tonight's explorations led me to the ✨gold✨ standard of mitigating the leakage of data in #ML -- #DifferentialPrivacy. The idea is to add very subtle statistical noise (in the dataset) to make it impossible to infer information about an individual data point.

Lately, I’ve been working on #PrivacyPreservingML 🔐 I got looped in some projects after Apple launched AppTrackingTransparency (ATT) framework, requiring iOS apps to ask permission to share users’ data w/ 3rd parties. This has triggered an industry-wide discussion on best practices to respect user privacy.

Volumes of crude data are available at our fingertips today, and the latest concept of a #DataLake helps store any type or volume of data as-is, process it in real-time or batch mode, and analyze it at scale 🤽🧵👇

DATA is the new oil🛢️ As #DataCentric approaches to #ML gather traction, access to diverse, comprehensive, and more importantly quality data has been the talk of the town. Along these lines, it's important to understand what does QUALITY really means in the context of DATA 🔢🧵👇🏻

Hey folks 👋🏻 For those who missed the talk by @AndrewYNg on #DataCentric approach to #MachineLearning, which aligns with our mission @DataForML, here is a quick recap 🧵👇🏻

Machine Learning models are as good as the data they consume🍴Data impacts performance, fairness, robustness & scalability of #ML Systems. If not taken care of, it leads to a TON of tech debt over time in a corporate setting, downstream effects of which are termed as DATA CASCADES 🌊 🧵👇🏻

I have been professionally working as a Machine Learning Engineer since more than 2 years now and also, recently co-authored a book titled “Sculpting Data for ML: The first act of Machine Learning”. My past few experience have taught me that data does not get its due limelight in #MachineLearning as compared to complex model architecture. Keeping up with 'more data beats clever algorithms, but better data beats more data', here are top 5 tips for polishing the dataset to effectively solve #ML problems 🤖👇🏻

2020 was unprecedented. My heart goes out to all who had a difficult year 🤍 Losing the privilege to socialize as much and be outdoors, it gave Rishabh Misra an opportunity to do something that was always on our bucket list.