lefnire/ocdevel

ocdevel's Issues

Math Primer PDF - Dead Link

"Math Primer PDF" resource is a dead link.

https://cs229.stanford.edu/syllabus.html -> 404 Not Found

Use https://cs229.stanford.edu/ for the Stanford cs229 course page. The link for the syllabus is a public google sheet that appears to get updated each term. There are many links to resources through this page, including a cheatsheet, a Matrix Cookbook, and a LinAlg review

Cheers!

Projects/experience "back of the card"

For the right side, do your magic on "flipping it" or "more info", whatever, highlighting a few key projects:

1. Habitica (icon, blurb)
2. Machine Learning Guide Podcast (same)
3. OCDevel projects (maybe just a list of the project titles from my github)

Machine learning for blockchain data

Hi,

My name is Daria. I’m reaching out to you on behalf of Neeraj Kashyap, the maintainer of Moonstream.to (GitHub: https://github.com/bugout-dev/moonstream). I think Neeraj could share an interesting story for your audience. He is a mathematician, and holds a Ph.D. in Number Theory. He has worked at Google on TensorFlow, and has built knowledge graphs that are being used actively by major US healthcare organizations.

Some topics that Neeraj would really love to discuss with you on your podcast, besides the one in the title:

1. Exploration and Discovery on Blockchains. (Neeraj did an article on this here)
2. Blockchain data for data scientists.
3. How to crawl blockchain data for analysis.

If there are any other topics that you think your audience would love to learn in discussion with Neeraj, we're open to suggestions.

Let me know if it's something you would be interested in.

OCDevel.com resources list discussion

Role v Podcast stronger distinction (left pane)

Currently looks like title / subtitle; I'd like a stronger distinction, (1) highlighting me as "machine learning engineer" for hire, then separately (2) here's a link to the podcast (in case that's what brought them to the website)

Blog: --fix-missing means cache cleans was done in docker, need apt-get update -y first

MLA: practical clustering with Kmeans, Agglomorative, DBSCAN, etc

• Discuss basics of hierarchical clustering vs cosine / Euclidean.
• Kmeans limited in dimension-size.
• Discuss FAISS, ANNOY, etc.
• Nclust via silhouette, inertia elbow + kneed, auto with dbscan/hdbscan
• Other clustering (spectral clustering, mean shift, affinity propagation, etc)

Use react-helmet for page headers

Podcast > "Become a sponsor"

I'm going to try to pull in podcast advertisers. I've seen some podcast have "become a sponsor" buttons which lead to an informational page on podcast stats, a contact form, etc. See what you can find

Podcast > Libsyn correct description field

Description field used different from: in my library per-episode v in-store per-episode

Podcast > Patreon button

See https://godotengine.org/article/godot-gets-brand-new-animation-editor-cinematic-support
Widget:

<a href="https://www.patreon.com/bePatron?u=6961428" data-patreon-widget-type="become-patron-button">Become a Patron!</a><script async src="https://c6.patreon.com/becomePatronButton.bundle.js"></script>

MLA: Docker

Traditional setup

• Raw: CUDA, cuDNN, pip/pipenv
• Anaconda: python versions + system requirements (CUDA/cuDNN)
• Cloud via EC2, GCP, Azure. Custom instance or AMI with dependencies. Sagemaker.

Docker

• Full stack setup, Windows via nvidia-docker on dev channel
• Cloud deploy via Batch, ECS, Kubernetes, etc

MLG: Cartesian Similarity

Social media Gnothi and email me a screenshot/link for 3-month access to Machine Learning Applied; commit code to the Github repository for life-access.

Normed distances link

• A norm is a function that assigns a strictly positive length to each vector in a vector space. link
• Minkowski is generalized. p_root(sum(xi-yi)^p). "p" = ? (1, 2, ..) for below.
• L1: Manhattan/city-block/taxicab. abs(x2-x1)+abs(y2-y1). Grid-like distance (triangle legs). Preferred for high-dim space.
• L2: Euclidean. sqrt((x2-x1)^2+(y2-y1)^2. sqrt(dot-product). Straight-line distance; min distance (Pythagorean triangle edge)
• Others: Mahalanobis, Chebyshev (p=inf), etc

Dot product

• A type of inner product.
  Outer-product: lies outside the involved planes. Inner-product: dot product lies inside the planes/axes involved link. Dot product: inner product on a finite dimensional Euclidean space link

Cosine (normalized dot)

Wrong Link

Multi-variable calculus link is wrong. Correct playlist link is here.

Consider podcasting software alternative to Audacity

Look into Reaper, Ardour.

Closing for now, Audacity works just fine.

MLA: FastAPI, Celery, Postgres

• Client: React -> Netlify, S3 + Cloudfront, etc
• Server: serverless if you can. See AWS Amplify (eg Lambda, Cognito, etc). Firebase (or Cloud Functions)
  • If need server, FastAPI - not Flask
• Postgres. Job server!
• ML server. Batch with Docker, SageMaker, Cortex, PaperSpace

Podcast on Spatial SQL

Spatial SQL is an essential part of the future of cloud native spatial analytics and something we think your listeners would really engage with. Our founder Javier de la Torre and VP of Solutions Engineering Matt Forrest are available whenever's convenient for you.

Since Spatial SQL uses the same elements and structure as regular SQL, it is compatible with all modern analytics stacks. Whether the use case is site selection for brick & mortar retail, route optimization in logistics, or customer segmentation in CPG - more and more businesses want to understand not only where things are happening, but also why they are happening there.

Podcast > Tags + legend

This will be more complex. You'll see in the code that I store my podcast content in a separate file machine-learning.js, in a hard-coded Markdown string, which gets parsed in the React code and turned into HTML. It's a bit wonky, but it lets me cleanly (a) generate the webpage, (b) convert to the format needed by other systems I use (namely Libsyn for hosting my podcast). Well in the Markdown strings (each episode.content), I have tags for the resources I recommend. Things like #easy #hard #video #book #audio etc. These are going to be pretty important to the users. What I want to do is make them pretty, more interactive, and connected to a legend describing the tag. Some options would be: have the different tags have different colors (but consistent, so #easy always has the same color). Hover over a tag expands it a bit, brings it more "into focus". Either a popover comes up wijth a description of that tag ("this resource is difficult, so save it for when you're well-caffeinated and only 20m/day"); or maybe there's a legend at the bottom of the page with tag=>description, and when you hover over a tag it highlights that section in the legend? Play around and see what you think.

Episode 3 resources

Hi,

I just started listening to the podcast and I'm really enjoying it so far. I just wanted to enquire about the resources for episode 3 on ocdevel.com, which I can't seem to view though I've tried different devices and browsers. I'm not sure if this is just an issue on my end but I thought I'd post an issue in case it isn't. Thank you again and I'm looking forward to listening to more episodes.

Move to Anchor

Free hosting, better integration with various services, simple advertising. Will cover Apple feed requirements that Libsyn not doing well.

• Migrate old rss.xml (waiting for deletion of someone who squatted my feed)
  • Ensure GUIDs & such still work, no duplication of content & destinations.
  • Esp. destinations which I setup manually? (Look into Libsyn>Settings)
• Redo first episode, mention Gnothi, Patreon, free access & ad-removal
• Add ads to eps (projects broken, so just import MP3s & insert)
  • Ad-free eps combed into Patreon
• Also look into Megaphone

See email "Podcasting with Anchor!" if anything needed

Duplicate episode mentioning MLA?

2 showing in my feed, delete the extra.