Category: Daily Thoughts

Attending an event and some focus time

Committed the first real post for nels.ai and have that system up and running for publishing on Wednesdays going forward.

I’m attending the CIO Future of Work Summit this morning. The first event was titled, “The Radical Next: Designing the Organization of the Future.” 

Throughout the day I stayed in the sessions. A few of them were pretty interesting:

  • “Is There an Easier Way to Manage and Remediate Risk?”
  • “Game Changer: Deploying Gen AI to Maximize Customer Experience”
  • “From Shadow IT to Shadow AI: How to Gain Visibility and Rein in Unauthorized Usage”
  • “Deliver a Cost-Effective, Sustainable & High-Quality Digital Workplace”
  • “Friend or Foe? What’s the Role of Automation and GenAI for the Future of Work?”

Edited and scheduled vlog day 33.

A lot of housekeeping tasks

Setup for the new nels.ai domain is now complete. The process ended up needing a bit of time (over a day) for the DNS to sort out. I’m going to spend some time focusing on building content for the new domain today. 

Polling for week 38 was completed and loaded. I need to do some benchmarking this week. 

I need to release a version of my base election prediction proxy model. The chalk model was not very predictive on a single factor. It’s time for a new and better model. 

I set up https://linktr.ee/nelslindahl today. This is one of those platforms that I’m not entirely sold on, but I thought it was time to build out as a profile. 

I signed up for https://openai.com/form/red-teaming-network

Working on learning

I set my alarm an hour early to join a class called “Application Development with Cloud Run” which was a part of Google’s Innovators Plus: Live Learning Events. I plan on attending a few of these events, but for some reason this series of events runs very early in the morning. 

You can see my Google Developer profile here: https://g.dev/nelslindahl 

Apparently I have a Google Cloud profile as well here: https://googlecloud.qwiklabs.com/public_profiles/ff63ebc9-8d5b-49a8-9c70-25c0c292ab73 

“LocalGPT Updates – Tips & Tricks”

Aligning the backlog for 2023

Today was the first day in the vlog series that I forgot to upload the video the day it was created and had to quickly finish the upload after my alarm went off. That meant instead of writing this delightful post I had to work on the upload flow. Don’t worry the video was posted an hour later in the day than the normal series, but it went live and all is well. I’m using a basic video editing workflow where all the content is created on my Pixel 7 Pro, content is edited in PowerDirector on the phone, and then it just gets uploaded to YouTube. I’m actually getting fairly proficient at using PowerDirector after 29 days of consecutive vlog updates. The YouTube Shorts format works well enough and making content that is always less than 60 seconds is a different sort of challenge. At this point, I think that is a better format than turning out sub 10 minute videos. 

Last night I got into a fight with my writing backlog. I have been looking at the remaining blocks of content for the year and trying to consider what could best occupy that space. I’m very clear on what will be created in 2024. I have a book in mind that needs to be written and I’ll go about creating it one block of content at a time. Probably from start to finish in my normal writing style. That effort will yield a roughly 20 chapter book. Aligning the backlog for 2023 has taken a bit of time and I’m really considering either combining my code development planning and writing planning or splitting them up. One of those two things needs to happen. Generally, I keep my writing backlog separate. That has worked out well enough. I’m starting to approach a window where I’m probably going to spend more time coding than writing words. I’m sure that is a good thing and naturally a part of the cycle of creation.

A moment of promise stood out

Yesterday was sort of a lost day of sorts. I know that some days are going to be less productive than others, but yesterday was a little bit on the disappointing side of things. Part of that is on me for not working toward starting the day with a reflection on what I accomplished the day before and considering the handful of things that would be best to accomplish today. Sure that sounds simplistic and to some degree being planful and working to deliver a defined backlog is a direct strategy. My backlog has about 200 blocks or more of content in it that need to be worked. My biggest push this week has been to really dive in and use LangChain both locally and within some notebooks [1]. That effort should translate to a few GitHub notebook outputs and maybe some fresh look items being added into the backlog.

I’m going to spend some time today learning a bit more about IDEFICS as well which Hugging Face just shared with the world [2]. 

Some time was spent working on the content blocks for weeks 139 to 142. 

Footnotes:

[1] https://www.langchain.com/[2] https://huggingface.co/blog/idefics

Some things are just routine

Some things are just routine. Uninterrupted writing during the earliest hours on Saturday and Sunday. Walking the dog in the morning. 25 evening push ups. Drinking a glass of water with every meal. These things just become a part of a regularly scheduled plan. It’s good to have some things that are routine. I’m really trying to focus on working on delivering the items in my writing backlog. Earlier this week things became a lot less stable. My reaction to that instability was not to power through a bunch of backlog items. It’s good to understand that reaction. I’m really just sitting down now to write a little bit of content. Everything within the foundation of my writing ability was shaken a little bit. Distinctly shaken, not stirred, is how things felt for the moment part. A whirlwind of emotion and reaction beyond anything that was going to just drive some writing. It was one of those moments where you have to take a step back and consider things within a more macro context. 

I’m actually thinking about releasing some type of poll on Twitter/X each day. That activity would be done just for my own personal amusement. I had considered doing something within the spaces (live audio). That idea never went anywhere. I have posted over 400 videos on YouTube. Apparently, I have posted as of this moment exactly 444 videos. A bunch of those happened during the time when I thought it would be fun to publish a daily vlog. Maybe going back and trying to do that again would be fun. The best way to do that would be to just commit to recording short videos on my Google Pixel 7 Pro, editing them using the PowerDirector application, and publishing them from the phone. That is pretty much how the last round of vlogging happened. I might string together a few random clips and do a test run without making it privately visible. We will see how that ends up going tomorrow.

You might have guessed that it is now the next day and I was able to run the publishing test on YouTube. Oddly enough, the system actually uploaded the video as my very first YouTube Shorts contribution. It was a pretty easy thing to accomplish and PowerDirector was just as easy to use as I remember from the last time around. Loading content to the project, editing it up a bit, and publishing took just a few minutes. It was a very low friction activity and it was sort of interesting to see it show up as a Shorts contribution. I think that happened due to the orientation of the video. All the little snippets I grabbed happened to be shot vertically. That may be the key to triggering that type of upload category. I’m not exactly sure on that one. 

I was able to get back into the routine of things for my Saturday morning early hour writing efforts. Two blocks of content were developed including weeks 139 and 141. I’ll have to pick back up with week 140 tomorrow and finish up editing the other two blocks of content. I may not be able to record any new podcast episodes this weekend. I did take the time to set up a podcast playlist on YouTube and I might go ahead and release the Friday August 25, 2023 edition in video format as a sort of test run of the process. Adding that type of video production to my workflow will change the publishing cost from the 10 minutes of audio creation to about 30 minutes per edition. If the video creation process becomes very involved, then things might jump up to 90 minutes per block of content. That could be worth it or it might be a thing to try out and fail faster on by adding it to the mythical stop doing list.