Week 1: Dev Journal of January
Learning
ruff: An amazing python code linter and formatter with very high performance, written in Rust.- Ruff GitHub: https://github.com/astral-sh/ruff
Docker: Containerise your application.- You can containerise multiple apps using
docker-compose. - There are many concepts in docker I have learned Docker Volumes, Port Expose, Environment Variables Initialisation. Left out concepts: Docker Networking, Docker Hub, etc.
- You can containerise multiple apps using
pyproject.toml: New way to configure you Python project.- I used this to configure my
ruffconfigs. - I have not completely learned this configs management file exclusively regarding python projects.
- I used this to configure my
- Because this is the first week of the year, I have reviewed my configuration files from GitHub repository and GitHub Gist.
Thoughts
- Ruff is amazing because it combines both linting and formatting together with its amazing performance speed ⚡.
- Docker makes your application very easy to setup for other developer.
- Learning MLOps is very iterative, disgusting, and many more. But when you learn it you will be like “WOW! What a service, product, concept.”
Week 2: Dev Journal of January
Learning
- VSCode UI Settings: Explore settings in VS Code IDE which will modify its UI colour (mainly).
- Discussed Cold Turkey Blocker issue on Windows: Windows have the path related issue. Repository | GH2
- Aggregation Pipeline in MongoDB: Used aggregation pipeline in my project which makes it easy to deal with data. I
don’t explicitly merging the datasets in the script.
- Commit where I did this in my yt-watch-history project.
Thoughts
- Creating ML system is very tedious. I am trying to understand how to integrate these amazing tools in this. ALthough, I am able to train and do prediction with my ML Model but I am not satisfied with my approach. Thats why I am learning new ways to design these systems.
Week 3: Dev Journal of January
Learning
mkdocs-material: I have this amazing documentation maker using.mdfiles in 1 day (almost). You can see my personal website hosted with GitHub Pages.- Freelance: @theabhinav002 queried to create #web-scrapping APIs for
real estate website.
- I have accepted the offer.
- Now, I am creating the web-scrapper for housing.com.
- yt-watch-history: I have been working around Channel
Recommendation System which recommend channels on the basis of channel’s videos title and tags. However, I have
completely build the model architecture but I wonder to put it in a pipeline to make it more concise to use and
define.
- I have discussed this system with @arpitpatel01.
Thoughts
- Always write your thoughts whichever things you are working on because it helps you to understand your thoughts and also helps you that how you explain your things or projects.
- While my freelancing process I feel like I am at my full potential and I’ve almost completed the project in 3-4 days.
I have written code to scrape housing.com. Also created APIs around it using
FastAPIframework.
Week 4: Dev Journal of January
Learning
yt-watch-history: This week I spend all the time on refactoring and understanding my project. It feels very difficult to work on but I made my mind and say that here I can push and pass my limits.- Custom Transformers using
sklearnAPI: I have not totally learned how to create a custom transformer usingsklearnAPI but I got the idea of it how to use it. Also, I have wrote some custom classes and tested it (it works well). - New Realease: I have released
v0.0.1ofyt-watch-historyproject.
Thoughts
- You can create your custom transformer using
sklearnand then put them into a pipeline which works like magic. - I am going to break my limit with this project (
yt-watch-history) because in this I am using too many new concepts like Docker, MLFlow, Customsklearntransformers (recently). Now, I am thinking about AWS 🤩.
Week 5: Dev Journal of January
Learning
learnwith.campusx.in: I have written a script which goes through the html of the website and using thecookiesit fetches the resources which were provided in the videos’ description.- Learning ML Concepts: I have learned how Gini Impurity/Index works by practicing it on paper with my hand and pencil. This helps me a lot and I (sort of) figured out, how should I learn things in ML.
- Blog on Decision Tree: I published a blog on decision tree on my website.
Thoughts
- Frustration: In these few days, I am not feeling that inspiration while doing anything. I feel frustrated and difficult to focus on things. It is like something which I can’t express but it is bad for me.