I build and write about AI, data science & privacy to see what will hold up beyond the hype cycles. Everything is open source and transparent.
Writing
Agentic Coding for Non-Vibe Coders
A workflow for building real things (not demos) with AI coding agents
The Dopamine Trap of AI Coding
The victory of solving a hard problem was sweeter when things used to take time. Now, getting a dopamine hit from watching an agent do all the work, is not the same at all.
Local AI matters, but it's not ready yet
Frontier models are extraordinary. On-device alternatives are not there yet for the average person. But the privacy and cost math is shifting fast.
What 4,887 AI prompts say about how we talk to machines
I analyzed a year of my Claude conversations. The patterns were more revealing than the content itself.
From game design to p-values: Anatomy of a real A/B experiment
Most data scientists inherit someone else's pipeline. I built one end-to-end — product design, instrumentation, statistics — to show how it's done at the highest levels.
Projects
A/B Testing Memory Game
A live experiment studying how game difficulty affects player behavior. Play the memory game, contribute real data, and explore the published statistical analysis.
How I Prompt 2025
Year-in-review visualization of your Claude conversations
Local LLM Bench
Benchmark local LLMs on structured data extraction accuracy
Open Source
@eeshansrivastava89Published Analysis
About & Recognition
I lead a data science team at Starbucks focused on AI/ML, personalization, and experimentation — including recommendation models powering the Starbucks app for 40M+ users.
12+ years across engineering, marketing analytics, modeling, and team leadership. CS undergrad, MBA. Not ex-FAANG. Not Ivy League. Just someone who likes making things work, learning by doing, and working with smart people.
On the side, I build apps on AI, data science & privacy to see what holds up beyond the hype cycles. Each project goes end-to-end — product design, engineering, analytics, real users, published analysis. Everything is open source and non-commercial.