Raman Hafiyatulin’s Post

Free Threading — Python's way to "goroutines", sort of. I’ve been experimenting with the new python3.14t builds. By combining Free-threading with AsyncIO, you can now: 1. Spawn worker threads (Parallelism). 2. Run an AsyncIO loop inside each (Concurrency). 3. Use queue.Queue as your "Channels" for thread-safe communication. The Result? True parallelism without the memory overhead of multiple processes. I’ve broken down the benchmarks and the "serialization tax" of subinterpreters in my latest write-up. If you're building high-scale backends, this is required reading. Read more: https://lnkd.in/gp3KBR_G #Python #DistributedSystems #Concurrency #Scalability #Performance

Dashboard with DASH. Using a series of synthetic data, I created this dashboard using the DASH library in Python, but I must admit that it's easier to work with the Shainy library in R. Which do you prefer? The advantages of using these two libraries I just mentioned are that they are completely free and that you can also share and interact with the image in a professional meeting without having to pay a single cent. This is much more expensive with other solutions. #DataVisualization #PythonProgramming #BusinessIntelligence

How rustystats makes use of Python & Rust to fit GLMs. Faster, less memory required, and more accurate*. I've had this conversation a few times over the last couple of weeks - it is quite common to use other languages for heavy computation within Python libraries. This is how I used Rust to make fitting GLMs as fast and memory efficient as I could. You can find the code here: https://lnkd.in/e4cR7gwc *based on the single conversion GLM I migrated from statsmodel that made use of ordered target encoding for categorical factors and a monotonic spline for one of the continuous factors.

I was tired of rewriting the same molecular fingerprint pipeline every time I had to process a large dataset. For 1M+ molecules, it usually means stitching together several scripts for joblib parallelism, saving chunks to avoid OOM, progress monitoring, keeping track of invalid molecules etc. Few months ago, I decided to turn that repeated setup into a small reusable python CLI: rdkit-fp PIP: pip install rdfp Github: https://lnkd.in/dpmQ9dU6 It’s an ordinary wrapper I built for myself, nothing original, but it might be useful for anyone doing large-scale fingerprint computation in Python/RDKit.

Newsflash: Python is the new Excel. Don't be the only one stuck with 1,048,576 rows. So to avoid this fate, here's a 7-day crash course to help you finally quit the green icon: https://lnkd.in/d7neSJXZ

For years, we accepted the GIL as a tax on Python performance. But with the "No-GIL" movement officially maturing in Python 3.14 and 3.15, we are finally unlocking true multi-core parallelism. It is a massive shift in how we think about CPU-bound tasks. We no longer have to default to multiprocessing and the memory overhead that comes with it just to bypass the lock. Seeing a single Python process actually saturate multiple cores without the "ceremony" of older workarounds feels like a new era for the language. The performance gap with Go or Rust is narrowing where it matters most, making Python an even stronger contender for high-throughput backends. Are you already experimenting with free-threaded builds for your heavy processing, or are you waiting for library support to catch up? #Python315 #PerformanceEngineering #BackendDevelopment #NoGIL #ProgrammingTrends

Understanding Division in Python – Integer vs Float While practicing Python, I worked on a simple yet important concept: division operations. a = int(input()) b = int(input()) print(a // b) print(a / b) What’s happening here? // → Integer Division Returns only the whole number (quotient), discarding decimals / → Float Division Returns the exact result, including decimal values -> Example: If the input is: 4 6 Output will be: 0 0.6666666666666666 -> Key Takeaway: Understanding the difference between integer and float division is essential when solving real-world problems, especially in data processing and algorithm design. #Python #CodingJourney #100DaysOfCode #LearningPython #ProgrammingBasics

Day 06/50 – LeetCode Solved Increasing Triplet Subsequence (#334) Learned how a simple greedy approach can replace brute force: Track two values (first, second) and detect a third larger number. Result: O(n) time | O(1) space Key takeaway: Smart tracking > checking all combinations Dhanraj Sahu #Day06 #LeetCode #DSA #Python #CodingJourney

Fun python 1/100 Ever wondered why these two give different outputs ? Its called Singleton in python …………….. ids = {id(True) for i in range(500)} print(ids) ids = {id(int("100")) for i in range(500)} print(ids) ……………… What you think would be the answer to each loop?

🧠 Python Trap You’ll Probably Hit Once When you create a list using multiplication like * 3, Python doesn’t create separate inner lists. Instead, it creates multiple references to the same list in memory. So when you modify one, all of them change together. But when you use a list comprehension, each inner list is created independently. That means changes stay isolated, exactly as you’d expect. This small difference is responsible for a lot of confusing bugs, especially in nested data structures. Reference: https://lnkd.in/gWBiknUH #pythonprogramming #learnpython #coding #python