Hi, this is Raj. You've chanced upon my digital note-box (~Zettelkasten) maintained in org-roam. I've circled over a lot of tools to maintain an efficient system in the past and have finally settled on a format that treats everything as text and allows me to spend more time within emacs. I find it difficult to keep track of all that I learn and this dump delegates the notion of remembering to a computer.

I like building connections between domains that are seemingly distinct: I'll be keeping it simple and this note-box will act as an intermediary for the all the blogs that I intend to regularly push out henceforth. I believe it should be simpler for me to improve my writing skills if I have an informal setup - other than my main blog - where I experiment with varying styles and do not worry about compatibility issues with my existing work.

A major proportion of the nodes in this web will be in a transient phase at all times - completeness is not a goal for each session : this is the input buffer. More polished versions will be posted on the main blog once a node(set) matures and is worth sharing.

I will be timestamping the notes so as to be able to keep track of how a node(set) evolves.



  • need to have a pruning session sometime - a lot of unnecessary nodes that could be better restructured as sub-trees.
  • some nodes are just plain useless and are an itch I scratched gone itchier.


  • an explicit :wip: tag denotes that I'm actively working on that node (different from :transient:)


  • the literate programming setup is live and working well with the publishing script.
  • I have it setup for rust, common lisp and elisp for now
  • mermaid-cli also works for flowcharts locally but need to figure out the publishing process for it to be rendered properly on the online buffer -> UPDATE (2 mins after I pushed this) : just needed to put in an exports: both in the src blocks..
  • making notes is second nature to me now -> whenever I'm studying/want to start a new project, I don't have to think much in terms of how I'll document the process. The snowball is rolling.
  • Noticing an significant decrease in the pre-thought of remembering stuff before I get start with a domain. Now that it is delegated to something with the memory of a steel trap, all that I learn is a few keystrokes away.
    • surprisingly, I have a mental parallel of this web that I implicitly call upon when remebering stuff -> that is an unexpected and beautiful side-effect…
  • Errors may creep in aynwehr -> I do'nt particurlayl edti what I wirte here and all is typed in a single pass, along with my thoughts.
    • this might result in needlessly long structures that could be better organized. Of course, I don't do this for the main blog. The buffer is not something that I maintain for perfectionism, but for maintaining a playful approach towards my studies.


Here are some conventions that I follow to organize my work:

  • Tagging
  • Timestamps
  • will be expressing math as S-expressions rather than latex
    • the former flow fluidly off my fingers
    • the latter I'll reserve when for publishing on the main blog: latex isn't exactly good while note taking - is a distraction.
    • my goal being encoding the concept (procedure/definition/relation) in a way that is more accessible to that of a programmer's mind
    • I also choose to do so as it'll help me learn map concepts from different domains into a common symbolic representation (hmmmm … : yeah…., math already does that - I guess I'm just being peevish)
  • A top-down knowledge building style : the human mind doesn't remember what it knows. Trying to remember a specific latent knowledge node in this graph will be more efficient than building all the base nodes one by one until I reach a practical level of application. I will add atoms as and when I see fit : I might even insert most from memory.
  • each major node has its own idea buffer to keep note of specific ideas that I haven't organized yet.
  • at any given point of time, several emtpy nodes may exist : I'm not trying to be exhaustive right away when exploring a domain. This is supposed to be an eternal work in progress. However, to indicate that I intend to work on a node later on and it's not just an empty atom, I've placed :tbp: (to be populated) tags on such nodes.

Tag list

  • meta
  • tool
  • tbp
  • ai
  • programming
  • trait
  • nlp
  • math
  • data
  • idealogy
  • python
  • book
  • linguistics
  • representation
  • lisp
  • cache
  • atomic
  • task
  • problem
  • posted
  • blog
  • skills
  • cv
  • root
  • rust
  • web
  • project
  • yt
  • nlu
  • java
  • tools
  • cs
  • transient
  • ml
  • writing
  • wip
Tags:None | Copyright (C) 2023 by Raj Patil