Notes from the
edge of the model.
Field notes on what actually breaks in production — agents, retrieval, evaluation, MLOps, and the career decisions nobody writes down. Longer arguments become papers; these are the rest.
354 posts
Training GPT-4 reportedly consumed enough energy to power thousands of homes for a year.
The carbon footprint of AI is real and growing. And as engineers, we can make choices that reduce it without sacrificing capability. Practical steps for more sustainable AI: Choose…
By Pranay Mahendrakar Read postAfter 250+ AI systems, 3 books, 7 papers, and a programming language built from scratch — here's…
I know that the fundamentals matter more than the latest framework. Math, algorithms, and system design don't go out of date. I know that data quality beats model complexity in 90% of…
By Pranay Mahendrakar Read postThe Phoenix programming language taught me something counterintuitive about innovation.
I didn't build it because the world needed another programming language. I built it because I wanted to understand how programming languages work from the inside. And that understanding…
By Pranay Mahendrakar Read post200 posts later, I'm even more convinced: the best time to start building AI is today.
Not because the market is hot (it is). Not because the salaries are good (they are). But because we're at a moment in history where a single engineer with a laptop can build things that…
By Pranay Mahendrakar Read postTo everyone who's been following this journey — thank you.
Every comment that said "this helped me in my interview." Every DM asking for advice that I was actually able to give. Every message from someone who started their AI career partly…
By Pranay Mahendrakar Read postPeople ask me what I'm building next.
The projects I'm exploring: a more sophisticated AI agent framework that handles real-world messiness better than current tools. An open source evaluation toolkit specifically for RAG…
By Pranay Mahendrakar Read post