43 karma · joined June 17, 2015
However, out of nowhere during the end of 2019 I was approached by a startup here which works in Document AI space. I am the only senior. I am juggling through research + engineering ML. But in some ways, I didn't anticipate that I would be enjoying so much here. But still, imposter syndrome hits hard often.
If I hadn't quit my initial startup plan, I don't know what might have happened to my research and career. I might have been more miserable.
:) ---
[0] - https://www.one-tab.com/
On the contrary, I do believe that using emojis in certain contexts are better. For instance, when you are euphoric and someone sends you a "happy-like" emoji, it makes you feel more "connected".
I guess, it's just me...
It's working fine on FT.
[0] - https://docsumo.com/free-tools/extract-tables-from-pdf-image...
It can be easily set up...
Up 3 will put me 3 directories behind the current one.
---
(Say, "panim" name might be a gimmick?)
--- [0] - https://github.com/NISH1001/panim
Now, we've moved onto to ML-based approach to train generic models that can be applied to variety of documents for table structure recognition.
[0] - https://docsumo.com/free-tools/extract-tables-from-pdf-image...
What I have found is, it just hasn't worked so far in terms of discoverability and search. So far, GitLab wikis have been seamless. Archival is fine with them, but search and knowledge re-assessment has been not so easy. My ideal solution (of course, haven't done so far in our company) is to have Knowledge management system similar to Roam/Obsidian/Logseq/Athens/TiddlyWiki. But we haven't figured out any mechanisms to host Obsidian. I've been personally putting the knowledge to Roam/Obsidian graphs. Even if we figured out to host, probably team collaboration would be pain in the ass.
So far, we're dumping everything to tara.ai and GitLab wikis, hoping that someday we might be able to index these data and then combine everything for search.
This also has been a topic with few of my friends for a long time now. Knowledge management at startup/corporate level is fragmented it seems. If we had tools to self-host networked-thoughts, it'd be great.
One question I have is: how much compatible will it be eventually with other tools like Roam/Obsidian/Logseq (when we want to import the graphs)?
It's been more than a year since I quit Twitter and the likes. I can't stress enough how good it feels to not have constant stream of mental hazard. I decided to rely on my whim to collect information using my own filter bubble (something I like to term as "information mindfulness"). I feel: as digital platforms are growing and evolving, with everyone inside the echo chambers [0], it's better to practice this "information mindfulness".
I am sure it depends on your own usecase for social media like Twitter; how you're going to sink in the platform. For me it didn't work. I hit my mental limit (finally after a long time).
[0] - At this stage, almost everything feels like an echo chamber.
This reminds me of write-up by Morgan Housel: long-term knowledge vs expiring knowledge[0]. To optimize that trade-off, I generally try to review, on a weekly basis, what I consume from the internet-verse. This has helped me in some ways. Not sure if it's gonna work or not for other people... Plus, moving the scale towards consuming long-forms is also helping me out. (I guess it depends on what type of bubble you are wrapping yourself into. For instance, only visiting specific sub-reddit and LW topics intentionally has been advantageous to me...)...
[0] - https://www.collaborativefund.com/blog/expiring-vs-lt-knowle...