Welcome to In My Lightcone
I document the things I’m learning here, mostly for my own reference. My curiosity jumps around a lot, so this ends up being a pretty eclectic mix of topics.
The name comes from physics — a lightcone contains all events that can influence you and what you can influence. This blog is basically my intellectual lightcone: the ideas I’m grappling with and the questions that won’t leave me alone.
What I’m figuring out
Right now I’m deep in a bunch of different areas that probably don’t seem related but somehow all scratch the same itch:
Optimal play in games - Understanding how to find the best strategies across different types of games: deterministic vs stochastic, complete vs incomplete information. This will involve learning game theory fundamentals, depending on the game structure.
Price discovery and auction design - How do individuals behave with limited information and cognitive biases in auctions? How does auction design influence whether the final transaction value reflects actual value? This pulls me into mechanism design, behavioral economics, and understanding how market structure affects outcomes.
Recommendation systems - I’m particularly curious about how these systems create user bubbles and whether there are ways to resolve this inherent property without destroying the user experience.
Experimentation - Bayesian approaches in experimentation and when they’re a useful addition to your toolkit as a data scientist.
Memory in LLMs - Knowledge graphs, graph databases, and ways to store and retrieve memory for AI agents in a cheap, quick and efficient way. I want to learn about semantic search, knowledge graphs, and other memory architectures.
These topics will definitely shift over time. That’s kind of the point.
Why write publicly?
Partly because writing forces me to actually understand things instead of just thinking I do. Working in industry forces you to be quick and efficient in your work which has it’s own merits. But you don’t get a lot of time to grind slowly. I want to use this blog to bring back those muscles that took years of training to develop. But also because I’m hoping to find people who are curious about similar stuff. If you’re working on any of these problems, or if something I write reminds you of something completely different that you’re thinking about, I’d love to hear from you. Maybe you’re part of the pack I’m looking for.
Email: arunravishankar@gmail.com | LinkedIn: linkedin.com/in/arunravishankar
Recent Posts
Latest thoughts and explorations:
