Hi everyone,
I’m working on a project where I need to automatically detect and highlight areas with freckles and acne on facial images using Python.

Has anyone worked on something similar? I’m looking for suggestions on:

• Which libraries or models to use (e.g., OpenCV, Mediapipe, Deep Learning, etc.)

• Any pre-trained models or datasets for skin condition detection

• Tips on image preprocessing or labeling

Any help, ideas, or code references would be greatly appreciated.
Thanks in advance!

I’m early in my career and would really appreciate some guidance from experienced professionals in AI/ML.

For my final year project, I focused on ML engineering—training models, exploring architectures, and working hands-on with tools like PyTorch. However, in my current (first) job, the work is centered around building GenAI/LLM applications using APIs like OpenAI and Gemini. It’s more about integrating existing models than developing or training them.

While the work is interesting, I feel uncertain about my growth. I’m not gaining deep technical experience or strengthening my core ML skills. Long-term, I’d like to either:

• Pursue a job abroad (e.g., in Europe), or

• Apply for a master’s in AI/ML with scholarship support.

I’m currently torn between three options:

1. Continuing with LLM application development (e.g., agents, API-based tools),

2. Shifting toward core ML engineering (research, training models), or

3. Trying to balance both.

If anyone has faced a similar crossroads or has insights into which path offers better learning and international opportunities, I’d be grateful for your advice.

Thanks in advance!

Hey everyone! 👋

I’m working on a Python project using Spyder as my IDE, and I’ve run into a version issue with PySimpleGUI. The latest version, PySimpleGUI 5, has some changes that aren’t compatible with parts of my code or the behavior I’m expecting. So I want to stick with version 4 for now.

My question is:
How can I make sure that Spyder uses PySimpleGUI version 4 instead of automatically picking up version 5?

I’m basically looking for guidance on how to downgrade or lock the version inside Spyder’s environment so I can continue working without any issues. If anyone has done this before or knows the best way to manage package versions in Spyder, I’d really appreciate your help!

Thanks in advance

I’ve been exploring tools like Dask and Datashader for handling large-scale data in Python, especially for interactive exploration and visualization. They seem promising, but I’m wondering how reliable they are in real-world, production environments. Have you found these tools stable and efficient enough for serious workloads, or do you prefer more established ecosystems like Apache Spark for scalability and robustness?