Atul Rai

Building cool stuff, learning along the way, and sharing the ride.

[Satish 🇮🇳 @bombaylives]

Credit card interest in India is already astronomical 42% APR +on top of that, 18% GST is slapped on the interest amount, pushing the effective burden to nearly 50%!
​Why treat it taxable "service" when standard loan interest is exempt?
​@FinMinIndia @PMOIndia @RBI @GST_Council

[Atul Rai @HeyAtulRai]

The real product isn’t AI intelligence, it’s real-time constraints.

We obsess over "accuracy" but ignore "speed" under load. If your voice agent lags, it doesn't matter how smart it is.

The engineering reality: https://t.co/hCSCotZrom

[Satish 🇮🇳 @bombaylives]

Extremely disappointing experience with @Flipkart — this is the third time my order has been cancelled. Hope brands like @nothing @nothingindia @cmfbynothing @OnePlus_IN reconsider selling through Flipkart.

[Atul Rai @HeyAtulRai]

Mastering SOLID principles strengthens your code's foundation:

Single Responsibility: One reason to change.

Open/Closed: Extend without modifying.

Liskov: Subtypes must work as base types.

Interface Segregation: No unused methods.

Dependency Inversion: Rely on abstractions.

[Atul Rai @HeyAtulRai]

The Python GIL (Global Interpreter Lock) limits multi-threaded performance in CPython by allowing only one thread to execute Python bytecode at a time. Great for simplicity & memory safety, but a bottleneck for CPU-bound tasks.
Soln: Multiprocessing!

[Atul Rai @HeyAtulRai]

DBT (Data Build Tool) empowers data teams to transform raw data into clean, actionable datasets using SQL.

Use cases:
- Data transformation with modular SQL models
- Enforcing data quality tests
- Auto-generating docs
- Version-controlled workflows

#DBT #DataEngineering

[Atul Rai @HeyAtulRai]

Airflow DAGs (Directed Acyclic Graphs) are workflows with tasks executed in a defined order. Use them for ETL pipelines, data engineering, ML workflows, data monitoring, or DevOps tasks. If it’s automatable, Airflow can handle it!
For more info: https://t.co/8YjmnaVOA7 #Airflow

[Atul Rai @HeyAtulRai]

10/10 Summary:
Choose Pydantic for type-driven validation & APIs.
Choose attrs for lightweight, flexible data classes.
⚖️ Both are excellent but serve different needs.

[Atul Rai @HeyAtulRai]

1/10 Pydantic vs attrs: Two powerful tools for data modeling in Python, but each shines in different scenarios. Let’s break it down:

[Atul Rai @HeyAtulRai]

9/10 Ecosystem Integration:
Pydantic integrates beautifully with frameworks like FastAPI and SQLAlchemy.
attrs has broader use cases due to its simplicity and low dependencies.
#PythonTools #DataModeling

[Atul Rai @HeyAtulRai]

6/6 TL;DR: Server-side cursors in PostgreSQL let you efficiently process large datasets by fetching rows incrementally. They're a lifesaver for apps with limited memory or batch processing needs.

[Atul Rai @HeyAtulRai]

1/1 What is a Server-Side Cursor in PostgreSQL?
When querying large datasets, fetching all rows at once can overload your application’s memory.
👉 A server-side cursor lets you process rows incrementally, keeping memory usage low!

[Atul Rai @HeyAtulRai]

5/5 When NOT to use server-side cursors?

For small datasets, they add unnecessary complexity.
If you need the full dataset immediately for in-memory processing.