When my architect asked for the vision behind the house I'm building in India, I sketched a lighthouse. Five floors. One central column. The bottom floor was a gallery — open, social, collaborative. The top floor was a sealed chamber for uninterrupted deep work. I was low-key proud of
Waymo is objectively better than an average human driver. Data shows it reduces injury-causing crashes by over 80%. Yet, if an autonomous vehicle makes a single mistake, it's front-page news. We demand near-perfection from machines while accepting massive error rates from humans. And rightfully so. We should apply
Grateful to have had my poster featured at the 2026 Google Data Science Summit in Sunnyvale today. Autorater Context Enrichment tackles something I've been thinking about for a while: static prompts don't age well. Autoraters built on fixed grounding miss nuance as products evolve, and the
A few months ago, during a coffee chat, a very smart AI mentor at Google told me something: "The best way to think about these agents," she said,"is like a very smart intern. Give them enough context, define the workflow, and they will execute the code
In 2010, Tom Preston-Werner, co-founder of GitHub, wrote a short essay arguing that engineers should write the README before writing a single line of code. His reason was simple: A perfect implementation of the wrong specification is worthless. A few months ago, I was having an amazing brainstorming session with
I wanted to post this at least a month ago, but I’ve decided to give myself permission to be a bit tardy. I just completed my first year at Google in December. While I expected to grow technically, I didn't expect to grow even more as a
I'm on a brief vacation this week — my way of navigating the typical first-week-of-the-year hassle. I visited the planetarium with my family today, where the presenter opened the show with a question: "Today we are going to talk about dust. How many of you are interested?"
One of my former Data Science managers described the role of a Data Scientist as "Truth Seeking." He didn't mention "Truth bearing," and I appreciated the nuance, now more than ever. This used to conjure in me an image of the mythical Hamsa (swan)
Some of my most rewarding work has been navigating the intricate nuances of model evaluation with Dave Spiegel, my mentor and manager at Google. We’ve spent non-trivial amount of hours debating the merits of different metrics, questioning assumptions, and trying to build a framework for evaluation that is both
Whenever anyone mentions Multi-armed bandit algorithms, the synonymous term reckoned by most data scientists would be "explore-exploit". Pls let me know if that's what came to your mind too! This article is an attempt to introduce a new perspective to associate MAB algorithms with a different
It’s been a while since my last post on Navigating A/B-Seas , but I was prodding myself to be back, doing the thing I enjoy the most - writing. Over the weekend, I decided to roll up my sleeves and bring to life a small side project I’ve
In my experience with data professionals, sizing is often one of the most misunderstood or misapplied efforts. This is a bold accusation, but I come from a place of atonement rather than audacity. I have done sizing the wrong way for years until I learnt the simple recipe behind the
Retention & LTV
Estimating the Lifetime value (LTV) is indispensable in understanding the health of the product as well as in evaluating if the Cost of Acquiring Customer (CAC) or Cost of Retaining Customer (CRC) is justified. But many product organizations never calculate this value since this is viewed as an exercise in
Experimentation & A/B Testing
Is 10 and 11 significantly different from each other? What about 10,000 & 11,000. Just comparing counts is unusual in the world of experimentation, since we are often focused on comparing differences in means or differences in proportions. In many cases, this nudges us to create metrics that
The AI-First Data Scientist
Since Alan Turing's fascination behind the question, "Can machines think?" (from his novel paper on Computing Machinery and Intelligence in 1950) to teaching machines long term and short-term attention (from "Attention is all you need" paper in 2017), the best brains and the biggest
Experimentation & A/B Testing
Note : This is the third part of a series of articles discussing variance reduction techniques in A/B testing. Written in collaboration with Athulya Ganapathi Kandy The previous parts can be found here: * PART I : All strata are useful, but some strata are more useful than the others | LinkedIn * PART
Experimentation & A/B Testing
Note: This is the second part of a series of articles discussing variance reduction techniques in A/B testing. The first part can be found here: All strata are useful, but some strata are more useful than the others | LinkedIn When considering variance reduction techniques, there is a very simple
Experimentation & A/B Testing
Written in collaboration with Athulya Ganapathi Kandy NOTE : This is the first part of a series of posts discussing variance reduction techniques in A/B testing Introduction The biggest adversary an analytics professional encounters in their experimentation journey is (statistical) insignificance, which is the antithesis of what A/B tests
Experimentation & A/B Testing
When it comes to multiple test corrections in A/B tests, two methods stand out - Bonferroni Correction and Benjamini-Hochberg (B-H) Corrections. In this blog post, we will explore: * Commonly overlooked multiple testing scenarios * The difference between Bonferroni Correction and B-H Correction * Python Implementation for both methods Commonly Overlooked Multiple
Personal
If you want to convince a million homes to install solar panels on their rooftops to power their homes, how would you do it? Informing the household, of their solar potential would be the first thing in my hitlist. But first of all, why would anyone want to go solar.