Hi, I'm Shubham, a data scientist

I work on recommendations, ranking, and applied ML. I write about the gap between clean textbook ideas and messy production behaviour: how metrics fail, how model assumptions leak into product decisions, and how data shapes what models can learn.

Email LinkedIn X (Twitter) Medium

Latest Writing

All posts →

RecSys Dec 18, 2025 - 12 min

Why MAP and MRR are not a good choice for Search Ranking

Search ranking practitioners often use Mean Reciprocal Rank (MRR) and Mean Average Precision (MAP) to assess the quality of their rankings. In this post, we will discuss why MAP and MRR are bad for search ranking. We then look at two metrics that serve as...

LLMs Jul 13, 2025 - 7 min

Multi-Head vs Multi-Query vs Grouped Query Attention

The title may be long, but the explanation isn't. Since the original Transformer, researchers have continuously tweaked its architecture, aiming to cut inference costs. Today, we’ll compare Multi-Head vs Multi-Query vs Grouped Query Attention to understand...

Model Deep Dives Apr 5, 2025 - 13 min

CatBoost- Inner Workings and Optimizations

Gradient boosting is a cornerstone technique for modeling tabular data due to its speed and simplicity. It delivers great results without any fuss. When you look around you'll see multiple options like LightGBM, XGBoost, etc. CatBoost is one such variant. In...