WIR-021

Recurring commitments

Health

Task	Frequency	Status	Progress
Exercise 30+ minutes.	x3/week	✅	3/3
Wake before 08:00.	x5/week	✅	5/5
Meditate 3+ minutes.	x5/week	❌	1/5

Relationships

Task	Frequency	Status	Progress
Contact 2+ family members/day.	Daily	✅	7/7
Visit 1+ new place in town/week.	x1/week	❌	0/1

Research

Task	Frequency	Status	Progress
Submit a short conceptual post to the website; cover a topic in math/AI.	x1/week	❌	0/1
Complete 1+ deep work session.	x6/week	✅	6/6

Previous week’s action items

• 1. Complete daily WIR check-ins
  • At end of each day, post progress on all recurring commitments to blog
    • Wednesday
    • Thursday
• 2. Update personal constitution
  • Modify all failed goals for 03/08
    • Segment weekly AI/math blog post goal into smaller chunks
• 3. Complete 2+ research reports + add to blog
• 4. Complete backprop post + add to blog

Work completed

Date	DOTW	Work Completed
03-09-26	Monday	[LOG] WIR + correspondence [RSR] PIML-QPE
03-10-26	Tuesday	Storm chasin’ ⛈️
03-11-26	Wednesday	[RSR] Computational graphs blogpost.
03-12-26	Thursday	[PhD] Emails/interview scheduling. [RSR] PIML-QPE – QPE error correction w/ quantile-loss; mixed objective loss [LOG] Built some cool meta-cognition/productivity software.
03-13-26	Friday	[RSR] PIML-QPE – Re-derive accumulated MRMS radar fields
03-14-26	Saturday	[RSR] PIML-QPE – ML-bias correction experiments over ROI
03-15-26	Sunday	[RSR] PIML-QPE – Dataset expansion/QC dashboard

Reflections

Let’s start with an overview of what went well this week. Certain recurring goals (e.g., waking up early, regular exercise) are getting easier to adhere to; good habits are setting in. Since establishing a personal constitution, we’ve generally been better about staying in touch with friends and family, and deliberately exploring new locations around town.

But, we still have ample room for improvement. We’ve failed to post updates to the ML-concepts series for two weeks now. Additionally, there are adulting-tasks that have been left on the backburner. My working theory is that when un-fun responsibilities lack a set time and place, the natural tendency is to avoid them. Another core issue has been an absence of weekly and monthly research goals. Last year, these targets helped us to deliver results (projects, papers, etc.) instead of spinning our wheels in perpetuity.

Concretely, we’ll divide action items for next week into three buckets:

• (1) Research
  • Aiming to improve activity as a collaborator, impose reasonable project targets.
• (2) Adulting
  • Need to put wheels in motion/correspond to prepare F26.
  • Appointments and whatnot must be handled as well.
• (3) Self-Improvement
  • Will set a target to read X-amount of “La Casa de los Espíritus”.

Action items

• Research
  • Send 5+ research/professional emails throughout the week.
  • [PIML-QPE]
    • Message X&Y updates
    • Re-calculate uncertainty figures w/ accumulated MRMS fields; update slides.
    • Show inclusion of spatial context beats MRMS radar QPE baseline on val set/case studies.
    • Project update (of any length) blog post.
• Adulting
  • (Tuesday) Call X to send Y to Z
  • (Monday/Tuesday) Email X @ Y
• Self-Improvement
  • Post 2+ daily-updates to the WIR
  • Read 10+ pages of “La Casa de los Espíritus”

Until next week 👋