# What did I work on in January?

I think it's time to document everything work-related. I am not sure if I can add the links to my work here, I’ll ask my boss.

#### What happened in January?

I primarily worked on two major tasks:

1. Deploying a **React-Django app** (which I built end to end) to production.
    
2. Working on the backend of an application that uses **LLM as a reasoning layer**.
    

Additionally, I worked on fetching reviews from third-party APIs and storing them in the database. Sounds simple, right? It actually is—but only if you handle edge cases properly, avoid rate limits, and ensure that unnecessary costs aren’t incurred.

#### Search Functionality in the React-Django App

I implemented a search feature using **PostgreSQL’s trigram search**, which comes as a built-in extension. Why trigram? Because the dataset was limited to just **90 records** - there was no need to overcomplicate things. Maybe in the future, I'll consider **Elasticsearch**.

#### What Did I Learn?

* **Dealing with clients**
    
* **Writing better documentation**
    
* **Deployments (those pesky config issues)**
    
* **Frontend adjustments**: minor changes, centering elements, aligning components—patience is key!
    

For serving static assets, I used **Apache**, while the Django app ran with `runserver` since we weren’t expecting much traffic.

#### Anything Novel This Month?

Honestly, nothing groundbreaking. I didn’t explore any new technologies - just relied on good old **reliable tech**. Maybe I should get back to my **Rust learning journey**. I also want to write systems level code, some kernel functions - given the AI boom - I think I can learn things easily. Some cuda, etc.

#### AI in Development

I used a lot of AI-generated code:

* **Claude** was useful for frontend work.
    
* **GPT-4o & 4o-mini** helped with Python code.
    
* **O1** assisted with deployment issues.
    

Research was mostly done via **Google**. Documentation is underrated—it really helps.

I enjoy working on **early-stage products**.

AI has definitely made coding **faster**, but debugging AI-generated errors can be time-consuming. Speaking of which, **o3-mini** just launched - I haven’t tried it yet.

**Adding some links I found useful**

* [https://www.postgresql.org/docs/current/pgtrgm.html](https://www.postgresql.org/docs/current/pgtrgm.html)
    
* [https://docs.djangoproject.com/en/4.1/ref/contrib/postgres/lookups/#trigram-similarity](https://docs.djangoproject.com/en/4.1/ref/contrib/postgres/lookups/#trigram-similarity)
    
* [https://claude.ai](https://claude.ai)
    
* [https://www.digitalocean.com/community/tutorials/how-to-configure-the-apache-web-server-on-an-ubuntu-or-debian-vps](https://www.digitalocean.com/community/tutorials/how-to-configure-the-apache-web-server-on-an-ubuntu-or-debian-vps)
    
* [https://whitenoise.readthedocs.io/en/stable/django.html](https://whitenoise.readthedocs.io/en/stable/django.html)