OKR - 2024 Review and 2025 Planning
This was the first year in my life where I felt a desperation to make the most of my time. I had far too many things I wanted to get better at (ML, math and physics, growing soft skills, travelling, readingā¦), but I realised that the time itself is the bottleneck to everything.
Therefore I set a personal OKR for the year: the āthree Rsā ā read, record, run ā as an attempt to choose three areas that I choose to focus on from everything I wanted to do. To my surprise, these OKRs had an overarching influence on my life this yearāfor once, my start-of-the-year goals didnāt get abandoned after the first few months. I guess being desperate really does make a difference š.
Hereās a breakdown of how those OKRs went, color-coded by status.
| Status | Color Code |
|---|---|
| Sent it | š¢ |
| Satisfactory | š” |
| Could have been better | š |
| Absolute Failure | š“ |
š”Ā Objective 1: Read
Key Result 1: Read 1 book a month
Status: š¢
I initially set the goal of reading 30 books this year, but scaled it down to 12 books a year earlier. But I did end up reading 30 books!
This section warrants a blog post on its own:
Key Result 2: Read 100 papers
Status: š”
I read ~50 papers that I tracked through remarkable and notion, and there are quite more that I skimmed through that werenāt tracked. I wouldnāt say that they would add up to 100, but Iām confident it would have been somewhere around 70.
This KR gave me two skills. First, I learned how to read papers more efficiently. To me, efficiency means finding the balance between sweeping too many knowledge holes under the rug and obsessing over every tiny detail to the point where I can't make progress. I learned the context of the broader topic (and improved how to get enough context if I donāt already have it) to decide whatās worth focusing on and what can stay in the back of my mind. Second, I discovered the environment where Iām most efficient at reading papers. Reading on my laptop? Abysmal, especially my work laptop, where I have a hundred other things I can attend to. Printing them out? Works well, but I donāt want to kill more trees or already-non-existent free space in my flat. Reading on remarkable on the other hand became a nice middle ground
Still, something about this KR leaves me uneasy.
Midway through the year, I found myself questioning the definition of reading a paper. If I skim the abstract and methods section and feel vaguely confident that I āget the idea,ā does that count? Do I need to power through the whole thing, appendix and all? Should I try deriving the math behind the idea?
I also started doubting if I was actually learning these topics as deeply as I wanted to. Watching the number of papers pile up in my DONE folder gave me a sense of accomplishment at first, but eventually I started wondering: Do I really get it?
For some people reading many papers might be the most efficient way to learn about topics that they care. It just might be something about the way I learn. For example, if someone asks me to imagine a loss surface with certain characteristics, Iāll struggle. I need to code it, visualise it, and āwalk around itā myself to fully grasp the concept. And if Iām honest, there are only a handful of papers Iāve read this year that clicked and stayed with me ā where I could confidently discuss the history, strengths, and limitations of the topic and implement something meaningful based on what I learned.
Key Result 3: Take 3 online courses
Status: š“
One could argue that this is not really reading and I agree. This goal stemmed from reading the papers ā I felt the need to understand the fundamentals better to build my knowledge on.
I wanted to learn thermodynamics (to understand the diffusion models better), information theory (to grow better intuition on uncertainty and probability), and ā¦ something else. I donāt even remember, and that shows how glaringly red this section should be.
I tried taking a thermodynamics course through MIT OpenCourseWare, but after a few weeks, I realised I need someone to hold me accountable to take this more seriously. I quickly gave up on this goal and didnāt even attempt it again.
š Ā Objective 2: Record
I'm terrible at organising the bits of my life that I might want to remember ā I donāt take many pictures, I donāt keep track of my thoughts or feelings in any way, and I donāt even bother with Instagram stories anymore! But I still love looking back at what Iāve done, thought, and felt, and I know Iāll regret it if I donāt leave myself something.
So, if recording things is mostly for me, why did I decide to include KRs that involve posting publicly? There are definitely more than enough content out there so whatever I produce, I bet there are better explained and more entertaining things. The sharing part publicly is more for me. Iāve spent too much time being scared of not being good enough or saying the wrong thing, so I end up keeping everything inside. The risk of not communicating at all feels way bigger than the risk of oversharing in my life. I want to get past that fear. Plus, I believe there are certain things I can only learn by producing something to share.
Write journal 3x a week.
Status: š
For some weeks, I binge-wrote the journal on Sundays. For some special weeks, I honestly wrote down my feelings and reflection throughout the week. At the beginning and towards the end of the year, the journals are non existent. I wrote about my day-to-day, the bible passages that I read, or simply a record of my feelings.
Although I was inconsistent in writing the journal everyday, one thing that stemmed out from this is a journal dedicated to my mom. My mom passed away from glioblastoma, stage IV brain cancer, four years ago. Up until this year, it was impossible for me to even talk about this to anyone else.
Still, at least once a dayāwhether itās the terrible consequences of that damned brain cancer, the guilt and regret over what I could have done better while she was sick, or simply memories of herāI think about her. This journal helped me jot down and make sense the thoughts and feelings that pop into my head.
In 2025 I want to invest in a system where I can record the fleeting thoughts better as forgetfulness of my own thoughts is the biggest bottleneck for the journaling.
Write a blog post at least once this year.
Status: š”
This ended up being a bit of a self reference, because this blog post is that one blog post that I wrote this year. š
Record playing bass and post it somewhere.
Status: š“
Nope, not done.
- I learned some new bass lines though!
š¢Ā Objective 3: Run
This goal is the one that Iām most proud of. I picked up running in last year November and began taking it seriously this year. Itās very surprising to me that I started running, because I hated and sucked at running as long as I remember.
By all means, I am an average beginner, but Iāve never been so happy about being average at something in my life. I discovered so many things through running:
- I love how often the state of the body dominates the state of mind. Countless sluggish, cranky mornings and brain-fried evenings have been fixed by runs as short as 3K. I gained a powerful option to exhaust before blaming the down turns on my mental state.
- I used to like what I am good at (and I think I still do), but running was one of the first few things that I learned to enjoy in long-term even if Iām not that good at it. My attitude towards the things that I am bad at, and therefore I wouldnāt like normally changed slightly through running.
- Watching myself improve over the year reshaped how I weigh the importance of talent in a success of something. Obviously the effort doesnāt always outweighs natural talent, but I learned that perseverance and a genuine enjoyment of the process can go a much longer way than what I thought before in making up for it.
Run a 5K in under 27 minutes.
Status: š¢
I ran two sub 27:00 5Ks and five sub 27:10 5Ks according to Strava āļø. Again, definitely not on the faster end, but Iām very proud of where I am compared to a year ago.
Do 25 park runs.
Status: š¢
I technically did 23 park runs in 2024 but the goal was to get that t-shirt with 25 park runs written on it! I finished the 25th park run a week before I left London for Neurips.
Run three races.
Status: š¢
I ran two half marathons, Hackney half and TromsĆø midnight run, and five Maverick trail runs. Plenty enough to cover three races.
The 2024 OKR was helpful in giving my year a sense of direction, but for 2025, I want to rethink how quantitative goals influence me. While having concrete targets like reading 12 books and 100 papers was motivating, I have to admit that I sometimes rushed to meet those numbers. I often chose materials that were easier to digest and skipped over ones that would have required more time and effort. As a result, I still feel like Iām missing the fundamentals ā reading deeply and fully understanding ideas to the point that I can play around with them.
Over the winter break, I set a few goals for myself:
- Review some talks from NeurIPS.
- Train some diffusion models from scratch, including:
- Setting up environments to use Apple GPUs.
- Reading papers related to the models I want to implement.
- Working out the math behind diffusion and seeing the different methods in effect.
- Read Gƶdel, Escher, Bach: an Eternal Golden Braid (_I got this book eight years ago and still havenāt finished i_t š).
Looking back, I think I subconsciously knew what I needed for 2025: goals that focus on depth and perseverance over quantity.
So with that, here is my 2025 OKR.
Objective 1: Immerse
I admit that āimmerseā doesnāt really tie these KRs together but I was determined to make my OKR to be IRL.
Read three classics.
Some candidates:
- GEB: I didnāt give up yet.
- A tale of two cities
- Romance of the Three Kingdoms (ģ¼źµģ§): I read the comic book version as a kid but I want to know how this book is as an adult.
Read any other books to give my brain a breather in between.
Log a hundred different āstreams of consciousnessā.
Hopefully journaling keeps improving this year without having to make it a KR. As I mentioned earlier, I want to invest in a system where I can log the spontaneous thoughts easier. The fruition of that system should be captured by this KR.
Objective 2: Run
I want to focus on uphill runs and longer runs.
- Run a 10k in 55 minutes.
- Run a trail run with less than 7 mins / km (Trail run with at least 15km and 500m elevation gain.)
Objective 3: Learn
Take at least one graduate-level physics or math course.
Some candidates:
- Thermodynamics / Statistical Mechanics
- Quantum mechanics
Play with three ML areas.
Read whichever papers and blog posts that help with implementation. Some candidates:
- Diffusion models: Continue from the work from winter break and add some variations.
- Audio Signal Processing
- Explainability
Write blog posts and make code available.