Vinay Danepalli
Software developer. Learning AI in public.
This is where I publish what I learn: notes and worked assignments, in case they save someone else time.
I wanted to be a scientist. I still spend too many nights looking up at the sky, and I still believe that with enough time and effort, anyone can learn anything.
01latest work
Fresh off the bench.
02currently exploring
My own constellation.
Five topics I keep coming back to, named the way astronomers name stars: α is the brightest.
-
Agentic AIPlanning, tool use, memory, and how agents fail.
-
LLM architectureFrom tokenization to inference.
-
Computer visionConvolutions to vision transformers.
-
Multimodal modelsText, image and audio in one model.
-
AI & societyThe economics: who gains, who pays.
03courses
Ground school.
- ERA V5 The School of AI · 5 assignments
- CME295: Transformers & LLMsStanford · no notes yet
- Agentic AIUdacity · no notes yet
- Data EngineeringDeepLearning.AI · no notes yet
04contact
Send a signal.
Replies usually arrive after dark.