Inside Out

Notes on seeking wisdom and crafting software

Year in review - 2023

This is our yearly review note. You’re reading the 2023 edition. Previous year’s note is here.

Scorecard

This was a good year from work and mental health perspective. Not so good from physical health in our family.

  • Reading: 15 papers, 12 books (goal: 23 books)
  • Writing: 24 posts (goal: 23 posts)
  • Work
    • Start of the year, we shipped a much awaited feature. Hockey stick growth and the feature was later proposed to be included in the “platform” instead of just one app.
    • ~80% of my team got a well deserved promotion. What a nice feeling! ❤️
    • Decided to move back to an IC role from brief stint in Management.
    • Spent a few months launching a startup idea inside company. Team delivered pretty well, and I ran a call center for a week ☎️
    • Learned some react, some redux and AI.
    • Shipped some ML models to production. Built chat systems, agents, and classification models. 5 AI experiments ☑️
    • And yes, starting this year every engineer is an AI engineer. Don’t ask me about data structures or algorithms anymore, please. Ask me my prompting strategy.
  • Home
    • Lab and projects
      • Several updates to this site: migrated from Gatsby to Astro, multiple version updates and one visual refresh.
      • Upgraded to a dual monitor setup.
      • Consolidated my reading workflow to a few tools.
      • Created a shared wiki and action plan for home (“when I am gone”).
      • Shipped multiple versions of spekt loggers, habito and noted. Spekt loggers are feature complete now.
    • Body and mind
      • Traveled a lot including a solo trip to the Himalayas after a decade.
      • We struggled this year with multiple health issues in my immediate family. See learnings below.
      • Weekly journal worked well. Reasonable progress on Philosophy.
      • Lagged severely on physical activity.

Lessons learned

  • Create an iteration or feedback loop for every project.
    • Prefer to iterate a lot with customers, much lesser with management and never solo.
    • When customers don’t exist, iterate with someone who can sell the feature to customers. For example, product management.
    • If none of the above exist, you’re working on a science project. Evaluate if that’s something worth the pursuit.
  • ”Build and they will come”. Corollary: “Build and they will fund”. We built, they never came, nor funded 🙁
  • Did I tell you there are platforms everywhere? Even between the platform and the app. Maybe it’s a sweet spot to work in.
  • AI’s inherent indeterminism is scary. Oldies like me need some more time to make peace with systems where the same action (input) now leads to different effects (output) for no apparent reason!
  • As a leader, creating a sense of urgency is not universally good. If the sense of urgency is “local”, i.e., no one else above you cares about it, you lose trust fast. And that’ll be the last time someone will stretch for you.
  • Once upon a time, there were two AI agent systems, doing exact same thing in two products. Do we need both? No, that’s DRY violation. Do we need both? Yes, both earn money for the company, and need to stick around.
  • The joy of letting go is indescribable. Let go of more and more things including news.
  • Being content is the ultimate treasure 1.
  • Show and tell is awesome. This year I blogged live while traveling. It is uplifting to pen down the thoughts while experiencing them.
  • My greatest vice this year was anger and impatience. I lost tons of mental peace with fruitless followups and panga with large consumer product providers. I must choose the right service providers, and my battles.
  • It took several months for our toddler to make the pedal go full-circle while cycling. And then one fine day he discovered the joy. It never stopped after that. Was there any value in me getting impatient? No.
  • Be intentional about health and understand the principles of each system of medicine. Evidence based methods are better if you’re short on time. Nature based methods are slow but work at the fundamental levels (lifestyle, food intake etc.). Both are complimentary.
  • Nature based medicine systems (e.g., Ayurvedic) were part of life at one time. Every grandmother used to know the common ailments and the treatments available in the kitchen. I verified this by taking a book (authored by an expert doctor) and validating the recommendations with common sense recommendations by my elders. It was amazing to see both match.
  • If I do not learn Ayurveda and related sciences, my next generation will never learn about those either. Trust in the system, principles and knowledge more than the practitioners.
  • I learned to ask for hypothesis for every medicine prescribed. This helps me rationalize whether the hypothesis was correct. It takes tons of time to do research but at the least I understand what I am taking.
  • Never do a thing today that you’re going to regret tomorrow 2.

Year ahead

  • Reading: 10 papers, 24 books.
  • Writing: 24 posts.
  • Work: 5 experiments.
  • Home
    • Lab and projects
      • 5 experiments with 2 AI/ML efforts.
      • Ship monthly. Ask for feedback.
    • Body and mind
      • Journal: weekly.
      • Walk: 3/week.
      • Philosophy: practice daily.

Footnotes

  1. We were taught this sloka as a teenager. It took several decades to understand the truth in it.
    सर्पाः पिवन्ति पवनं न च दुर्बलास्ते
    शुष्कैस्तृणैर्वनगजा बलिनो भवन्ति |
    कन्दैः फ़लैर्मुनिवरा: क्षपयन्ति कालं
    संतोष एव पुरुषस्य परं निधानम् ||
    Serpents drink air, but they are not weak.
    With dry grasses, wild elephants become strong.
    With roots and fruits, the best sages pass the time.
    For man, contentment alone is the supreme treasure.

  2. किमहं साधु नाकरवम्‌। किमहं पापमकरवमिति
    Why have I not done the good? Why have I done that was evil? Taittiriya Upanishad, Brahmananda Valli, 9.1.