new agenda

Author: Teinc3

markdowns/agenda/23062026.md

@@ -0,0 +1,37 @@
+## Daily Agenda
+*Date: July 23, 2026*
+
+Unfortunately I just don't have enough time. Have a look at what I have to simultaneously manage 
+literally in my holidays just to prove my commercial value a few months later:
+
+1. G2G Traversal Benchmark Harness. Aim to be doing a 3D Multi-modal run as PI pointed out it would
+be really useful for clinical evaluations and might even help get a pioneering paper published for
+GPT-5.6 performance benchmark with other models on the scope of medical oncology.
+2. CUDA programming tasks. The uni-sponsored course is about to end, but I still have some
+compute-intensive, non-memory-hard POW functions to break in Territorial.io. I don't have a GPU
+myself so I would still love to request persistent access to the htc cluster, but there seems to be
+a ton of paperwork for that. Or maybe there isn't and I just have to talk to the course PI, which
+I'll do tomorow.
+3. LC Grind, it's the final stretch so we'll need 4 questions per day. Currently doing hard SQL
+problems. I've actually realised how much SQL i've been using to query my databases for the Media 
+stuff and every time I do these SQL problems I'm just overengineering the syntax. At least now I got
+a cheat-sheet compiled so I actually get to know the syntax instead of spamming retries trying to
+figure out why WHERE doesn't work with aggregate functions.
+4. PT Media programme automation and management. I suppose that's fine once the creators settle in
+but atm it's just lots of work onboarding them and manually translating and explaining rules simply
+because they just don't understand the instructions at all. At least I get to improve my Rulang from
+doing so, I'm not really complaining. But I just don't think the administration sees the potential
+value that I can deliver. I understand it mostly comes down to trust, but the point is if it's
+purely about trust then you should probably be less reluctant to let me do these jobs considering 
+they probably know my full abilities and skillsets and yet my commitment to product safety.
+5. Rulang: 100 card reviews a day, 1-2 podcasts a day, bunch of passive immersion. At least the time
+spent on Anki is replacing the time spent on doomscrolling so it's definitely a good thing.
+
+
+The grind is real, but I really hope I can stop in October if I am lucky enough to obtain an FDE
+intern offer. But in all fairness, for my senior wrangler friend out here, it's just the same grind
+for them, just more academic instead of commercial/industrial. Even though their immediate ROI can
+be as much as 6x that of mine, thanks to AI/HFT poachers offering somewhat close to a 600k starting
+salary for newgrads. Another IMO friend with perfect GPA at MIT also landed a research position in
+OpenAI, so probably somewhere even higher than 600k since they now live in America. Absolutely crazy
+to think about, although my end goal was never about money and more about leverage really.

markdowns/thoughts/21072026.md

@@ -1,4 +1,4 @@
-Previously I used to scoff on people for enforcing 80-character line limits. I've always never cared
+Previously I used to scoff at people for enforcing 80-character line limits. I've always never cared
 about it until I decided to put useful information on the right panel (cough cough AI) pushing me to
 a 100-character line limit. But take a look about this unmatched information density on 80: