Parcourir les catégories
Explorer
Fiverr Pro
Français
$
USD
I will review, debug, and improve your python machine learning code
Senior Data Scientist
I am a Data Scientist with an interest in Bioinformatics and Machine Learning. I have a strong understanding of software and ML model development, backed by over 10 years of programming experience in ...
À propos de ce service
Are your machine learning models underperforming, throwing errors, or just hard to understand? I'll help you fix that.
I'm a Senior Data Scientist at a Bank with 3+ years of production ML experience, a First Class CS degree, and I am currently completing an MSc in Bioinformatics. I've built, debugged, and optimised real ML models at scale, not just toy projects.
What I can review and fix:
- Scikit-learn and PyTorch pipelines and models
- Pandas data preprocessing and feature engineering
- Model training, evaluation and hyperparameter logic
- Code structure, efficiency and best practices
- NumPy, XGBoost, PyTorch and general Python ML code
What you'll get back:
- Corrected, clean Python scripts
- A written report explaining every issue found and the fix applied
- Optimisation suggestions to improve speed and accuracy
- A custom evaluation script outputting key metrics
- Model documentation covering inputs, outputs, assumptions and limitations
To get started, send me:
- Your Python script(s) as .py or .ipynb files
- A brief description of what the code is supposed to do
- Any error messages or specific problems you've noticed
I respond to all messages within 24 hours. Let's get your code working properly.
Langage de programmation:
Python
Frameworks:
Scikit-learn
•
PyTorch
•
Panda
Outils:
Jupyter Notebook
•
tensorflow
•
Excel
Mon portfolio
FAQ
What types of machine learning problems can you help with?
I can help with classification, regression, clustering, and neural network models. I work with Scikit-learn, PyTorch, XGBoost, and general Python ML code using Pandas and NumPy. If you're unsure whether your project fits, message me before ordering.
What file formats do you accept?
I accept .py Python scripts and .ipynb Jupyter notebooks. If your project has multiple files, zip them and send as a single archive. Please also include any CSV or data files the code depends on, or a small sample of them.
What if I don't know what's wrong with my code, I just know it isn't working?
That's completely fine and, in fact, the most common situation. Just describe what the code is supposed to do and what's happening instead. I'll diagnose the issue as part of the review.
Will you share or use my code for anything else?
No. Your code is completely confidential. I only use it to complete your order and it is not shared, stored, or reused in any way.
How do I know which package to choose?
Choose Basic if you need a review and fix of 1-2 scripts. Choose Standard if you have up to 3 scripts and want optimisation on top of fixes. Choose Premium if you have a full project of up to 5 scripts and want an evaluation script and model documentation included.
What is the evaluation script included in Premium?
It's a standalone Python script tailored to your model that you can run after any retraining. It automatically loads your model, runs predictions on test data, and outputs key performance metrics. Results are also saved to a log file with a timestamp so you can track performance over time.
Do you guarantee my model will be more accurate after the review?
I can fix bugs, remove data leakage, improve preprocessing, and suggest better hyperparameters or algorithms, all of which typically improve performance. However, accuracy improvements depend heavily on your data quality and problem complexity, which I cannot guarantee.
What if I need changes after delivery?
Each package includes revisions: 1 for Basic, 2 for Standard, and 3 for Premium. Revisions cover adjustments to the work already delivered. If you need something outside the original scope, we can discuss it as a separate order.
How should I send my data if it's sensitive or large?
For sensitive data, send a anonymised or dummy sample that has the same structure and column names as your real data. This is enough for me to review and fix your code. I do not need your actual data to assess the code itself.
