Swastik Verma
HomeSkillsProjectsAchievementsContactArticles
Resume

Insights & Architecture

Technical Writing
& Research

Open to Collaboration

I regularly publish new articles and papers here. Have a Machine Learning research idea or want to brainstorm architecture?

Email Me

Why Linear Regression Minimizes Squared Error: A Probabilistic Explanation

Exploring the mathematical foundations of linear regression and diving deep into the probabilistic reasoning behind minimizing squared error.

Mar 2024 5 min read
Machine LearningMathematicsProbability

The Hidden Machinery Behind Modern AI: A Deep Dive into Data Preprocessing for Data Streams

Analyzing the critical data preprocessing pipelines required for handling continuous, real-time data streams in modern AI architectures.

Feb 2024 7 min read
AIData EngineeringPipelines

Structural Shift in Peak Power Demand — Delhi Grid (2020–2025)

A formal research paper (under review · ICIDEA) modeling peak-demand growth via a Heat×Humidity interaction term to capture a 'Cooling Trap' that standard load models miss. A RidgeCV regression on 12 years of grid data reaches R² 0.96 (train) / 0.93 (2024 hold-out) at RMSE 372 MW, quantifying nocturnal load divergence to motivate battery-storage deployment.

Under review · ICIDEA Academic Paper
RidgeCV RegressionEnergy ModelingUnder Review

Why Linear Regression Minimizes Squared Error: A Probabilistic Explanation

The Hidden Machinery Behind Modern AI: A Deep Dive into Data Preprocessing for Data Streams

Structural Shift in Peak Power Demand — Delhi Grid (2020–2025)