new questions from the launch stream

alexeygrigorev · alexeygrigorev · commit 219639f20956 · 2025-09-19T12:03:04.000+02:00
diff --git a/_questions/machine-learning-zoomcamp/general/001_0e38656cfb_how-do-i-submit-homework.md b/_questions/machine-learning-zoomcamp/general/001_0e38656cfb_how-do-i-submit-homework.md
@@ -7,6 +7,6 @@ sort_order: 1
 - Do the tasks locally
 - Publish your code (e.g., in your own GitHub repo)
 - Submit your answers via the homework form and include the URL to your code
-- Answers are revealed after the deadline closes
-- Homeworks are in the cohorts folder, e.g. for 2025 it's cohorts/2025 (https://github.com/DataTalksClub/machine-learning-zoomcamp/tree/master/cohorts/2025)
-- The forms for submitting the homework are in the course management platform: https://courses.datatalks.club/
+- You will see the answers only after the deadline
+- Homeworks are in the cohorts folder, e.g. for 2025 it's [`cohorts/2025`](https://github.com/DataTalksClub/machine-learning-zoomcamp/tree/master/cohorts/2025)
+- The forms for submitting the homework are in the [course management platform](https://courses.datatalks.club/)
diff --git a/_questions/machine-learning-zoomcamp/general/001_4d5aa45b03_are-jupyter-notebooks-used.md b/_questions/machine-learning-zoomcamp/general/001_4d5aa45b03_are-jupyter-notebooks-used.md
@@ -0,0 +1,7 @@
+---
+id: 4d5aa45b03
+question: Are Jupyter Notebooks used?
+sort_order: 1
+---
+
+Yes. You’ll work extensively with notebooks alongside standard Python files and CLI tools.
diff --git a/_questions/machine-learning-zoomcamp/general/001_f9c9e0b9eb_updates-fastapi-pytorch-repo.md b/_questions/machine-learning-zoomcamp/general/001_f9c9e0b9eb_updates-fastapi-pytorch-repo.md
@@ -0,0 +1,7 @@
+---
+id: f9c9e0b9eb
+question: Will the repo already have all FastAPI/uv/PyTorch updates?
+sort_order: 1
+---
+
+The repository does not lock in every update at once. Some updates for FastAPI/uv and PyTorch are released as the cohort progresses. For theory, older videos remain; for implementation, updated workshops and videos are linked (for example, the FastAPI+uv workshop and the PyTorch add-on). To find the latest materials, check the module pages for updated workshop links. If a critical update is pushed, instructors will annotate or replace the relevant material accordingly.
diff --git a/_questions/machine-learning-zoomcamp/general/002_8fd8cd336d_how-do-i-sign-up.md b/_questions/machine-learning-zoomcamp/general/002_8fd8cd336d_how-do-i-sign-up.md
@@ -1,7 +1,7 @@
 ---
 id: 8fd8cd336d
 question: How do I sign up?
-sort_order: 1
+sort_order: 2
 ---
 
 In the course [GitHub repository](http://mlzoomcamp.com), there’s a link to sign up. Here it is: [airtable.com](https://airtable.com/shryxwLd0COOEaqXo)
diff --git a/_questions/machine-learning-zoomcamp/general/002_9155c70324_when-is-the-next-cohort.md b/_questions/machine-learning-zoomcamp/general/002_9155c70324_when-is-the-next-cohort.md
@@ -4,7 +4,7 @@ question: When is the next cohort?
 sort_order: 2
 ---
 
-The course is available in a self-paced mode, allowing you to go through the materials anytime. If you prefer to join a cohort with other students, the next sessions will begin in September 2023, September 2024, and September 2025.
+The course is available in a self-paced mode, allowing you to go through the materials anytime. If you prefer to join a cohort with other students, live cohorts start every September.
 
 Zoomcamps are scheduled throughout the year. For more information, refer to [A Guide to Free Online Courses at DataTalks.Club](https://datatalks.club/blog/guide-to-free-online-courses-at-datatalks-club.html).
 
diff --git a/_questions/machine-learning-zoomcamp/general/002_9cf1c56e5b_should-i-star-the-github-repo.md b/_questions/machine-learning-zoomcamp/general/002_9cf1c56e5b_should-i-star-the-github-repo.md
@@ -1,7 +1,7 @@
 ---
 id: 9cf1c56e5b
 question: Should I star the GitHub repo?
-sort_order: 1
+sort_order: 2
 ---
 
 Yes, please star the repository: https://github.com/DataTalksClub/machine-learning-zoomcamp/. Starring helps it trend so others can discover the course, and the instructor explicitly asks attendees to star it to boost visibility.
diff --git a/_questions/machine-learning-zoomcamp/general/002_e0a95572a6_join-slack-invite-not-arrived.md b/_questions/machine-learning-zoomcamp/general/002_e0a95572a6_join-slack-invite-not-arrived.md
@@ -0,0 +1,7 @@
+---
+id: e0a95572a6
+question: How do I join Slack if the invite email didn’t arrive?
+sort_order: 2
+---
+
+Go to DataTalks.Club, request a Slack invite, or use the manual request form (processed daily). After joining, browse channels and join **#course-ml-zoomcamp**.
diff --git a/_questions/machine-learning-zoomcamp/general/003_1f17886bde_how-to-get-help-if-stuck.md b/_questions/machine-learning-zoomcamp/general/003_1f17886bde_how-to-get-help-if-stuck.md
@@ -0,0 +1,10 @@
+---
+id: 1f17886bde
+question: How do I get help if I’m stuck?
+sort_order: 3
+---
+
+- [Slack](https://datatalks.club/slack.html): paste your code and errors, and use threads to keep discussions organized.
+- this FAQ and [past office hours](https://www.youtube.com/@DataTalksClub/search?query=ml%20zoomcamp%20office%20hours): check these resources for guidance before asking.
+- AI assistants like ChatGPT: use them for quick guidance or brainstorming.
+- Peers and instructors: they may chime in when available.
diff --git a/_questions/machine-learning-zoomcamp/general/003_2b5ff70c77_no-enrollment-before-submit.md b/_questions/machine-learning-zoomcamp/general/003_2b5ff70c77_no-enrollment-before-submit.md
@@ -0,0 +1,7 @@
+---
+id: 2b5ff70c77
+question: Do I need to enroll in the course before submitting homework?
+sort_order: 3
+---
+
+No enrollment is required to submit homework. Just log into the homework form when it opens. The Airtable registration you may see is only for announcements; actual submissions are made on the course platform forms and via your GitHub as specified in the homework guidelines.
diff --git a/_questions/machine-learning-zoomcamp/general/004_0d854c74af_donations-sponsorships.md b/_questions/machine-learning-zoomcamp/general/004_0d854c74af_donations-sponsorships.md
@@ -0,0 +1,10 @@
+---
+id: 0d854c74af
+question: How do donations/sponsorships work?
+sort_order: 4
+---
+
+- The course is free.
+- Sponsors may be added.
+- Voluntary donations help too. You can donate here: https://github.com/sponsors/alexeygrigorev.
+- If you have a company training budget, you can request an invoice to support the course.
diff --git a/_questions/machine-learning-zoomcamp/general/005_90323db1ac_None.md b/_questions/machine-learning-zoomcamp/general/005_90323db1ac_None.md
@@ -0,0 +1,7 @@
+---
+id: 90323db1ac
+question: What’s the recent pass/completion rate?
+sort_order: 5
+---
+
+In the most recent cohort's dashboard snapshot, roughly 85% of project attempts passed.
diff --git a/_questions/machine-learning-zoomcamp/general/006_2453bf4760_spark-bigdata-tooling.md b/_questions/machine-learning-zoomcamp/general/006_2453bf4760_spark-bigdata-tooling.md
@@ -0,0 +1,7 @@
+---
+id: 2453bf4760
+question: Will we cover Spark or big-data tooling?
+sort_order: 6
+---
+
+No Spark in this course. The focus is on core ML engineering and deployment patterns. We do cover deployment tooling relevant to ML (e.g., Kubernetes, TensorFlow Serving, KServe) as part of the deployment modules; see Modules 5–11 for details.
diff --git a/_questions/machine-learning-zoomcamp/general/009_ef9594914c_how-to-get-answers-common-questions.md b/_questions/machine-learning-zoomcamp/general/009_ef9594914c_how-to-get-answers-common-questions.md
@@ -0,0 +1,7 @@
+---
+id: ef9594914c
+question: How can I get answers to common questions?
+sort_order: 9
+---
+
+Start with the FAQ document (general + per-module Q&A). There’s also a Zoomcamp Q&A bot in Slack—use it thoughtfully; often the FAQ or recent messages already contain your answer.
diff --git a/_questions/machine-learning-zoomcamp/general/010_62f5aa4d83_data-engineers-to-dsml.md b/_questions/machine-learning-zoomcamp/general/010_62f5aa4d83_data-engineers-to-dsml.md
@@ -0,0 +1,7 @@
+---
+id: 62f5aa4d83
+question: Can this program help data engineers move into DS/ML engineering?
+sort_order: 10
+---
+
+Yes. The program’s project-first flow, deployment modules (FastAPI/Lambda/Kubernetes), and evaluation practices map well to ML engineer roles. Practically, you’ll gain experience with end-to-end ML workflows—from data engineering pipelines and model deployment to evaluation and monitoring—using tools commonly used in DS/ML roles.
diff --git a/_questions/machine-learning-zoomcamp/general/014_c02a8e0913_how-announcements-work.md b/_questions/machine-learning-zoomcamp/general/014_c02a8e0913_how-announcements-work.md
@@ -0,0 +1,7 @@
+---
+id: c02a8e0913
+question: How do announcements work?
+sort_order: 14
+---
+
+Announcements are posted in the Telegram channel and mirrored into Slack. It’s best to join Telegram for clean, broadcast-only updates; otherwise you’ll still see mirrored posts in Slack.
diff --git a/_questions/machine-learning-zoomcamp/general/016_679b977c92_where-materials.md b/_questions/machine-learning-zoomcamp/general/016_679b977c92_where-materials.md
@@ -1,11 +1,14 @@
 ---
 id: 679b977c92
-question: Where do I find all course materials and links?
+question: Where can I find all course materials and how can I keep track of lessons
+  and modules?
 sort_order: 16
 ---
 
-- All course materials live in the main ML Zoomcamp GitHub repository: https://github.com/DataTalksClub/machine-learning-zoomcamp/
-- Modules are folders for each unit, with a README/markdown per unit, plus videos and notes.
-- Cohort-specific items (homework and dates) are under cohorts/2025/.
-- The YouTube playlist "Machine Learning Zoomcamp" contains all lessons.
-- For quick access, browse the repo, check the modules folders, and look in cohorts/2025 for cohort-specific items. The syllabus is also available at mlzoomcamp.com/#syllabus.
+All course materials and links live in the main ML Zoomcamp GitHub repository:
+- Repository: https://github.com/DataTalksClub/machine-learning-zoomcamp/
+- Modules: folders for each unit, with a README/markdown per unit, plus videos and notes.
+- Cohort-specific items (homework and dates) are under cohorts/2025.
+- YouTube playlist: https://www.youtube.com/playlist?list=PL3MmuxUbc_hIhxl5Ji8t4O6lPAOpHaCLR contains all lessons.
+- Year-specific playlists (2021–2025) contain cohort-specific streams and extras.
+- Quick access tips: browse the repo, check the modules folders, and look in cohorts/2025 for cohort-specific items. The syllabus is also available in the course repo mlzoomcamp.com/#syllabus.
diff --git a/_questions/machine-learning-zoomcamp/general/103_bc2e78caf3_slack-house-rules.md b/_questions/machine-learning-zoomcamp/general/103_bc2e78caf3_slack-house-rules.md
@@ -0,0 +1,12 @@
+---
+id: bc2e78caf3
+question: What are the Slack “house rules”?
+sort_order: 103
+---
+
+Slack house rules:
+- Ask course questions in `#course-ml-zoomcamp` (not `#general`).
+- Use threads to reply.
+- Paste text/code instead of screenshots or phone photos.
+- Don’t tag instructors; many peers can help and instructors see messages anyway.
+- Keep the channel tidy and on-topic.
diff --git a/_questions/machine-learning-zoomcamp/general/104_3b78e09b80_uv-anaconda-env-tools.md b/_questions/machine-learning-zoomcamp/general/104_3b78e09b80_uv-anaconda-env-tools.md
@@ -0,0 +1,9 @@
+---
+id: 3b78e09b80
+question: 'Which tools do we use for environments: uv or Anaconda?'
+sort_order: 104
+---
+
+- Install Python easily with Anaconda (especially on Windows).
+- Use uv for virtual environments and package installs (recommended over conda for this course).
+- The uv + FastAPI workshop is included in Module 5.
diff --git a/_questions/machine-learning-zoomcamp/general/105_0953e15dc6_cloud-provider-used-aws.md b/_questions/machine-learning-zoomcamp/general/105_0953e15dc6_cloud-provider-used-aws.md
@@ -0,0 +1,7 @@
+---
+id: 0953e15dc6
+question: Which cloud provider is used?
+sort_order: 105
+---
+
+Examples use AWS (you should have or create an AWS account). Concepts transfer to other clouds (GCP/Azure) with minor adjustments.
diff --git a/_questions/machine-learning-zoomcamp/general/106_9198d40c17_system-design-included.md b/_questions/machine-learning-zoomcamp/general/106_9198d40c17_system-design-included.md
@@ -0,0 +1,7 @@
+---
+id: 9198d40c17
+question: Is system design included?
+sort_order: 106
+---
+
+There is no dedicated system-design module. The course emphasizes end-to-end ML projects and deployments (web services, Lambda, Kubernetes). You will learn by building and deploying models rather than focusing on standalone system-design concepts.
diff --git a/_questions/machine-learning-zoomcamp/general/107_9ffc454a24_leaderboard-advantage.md b/_questions/machine-learning-zoomcamp/general/107_9ffc454a24_leaderboard-advantage.md
@@ -0,0 +1,7 @@
+---
+id: 9ffc454a24
+question: Is there any advantage to high homework/leaderboard scores?
+sort_order: 107
+---
+
+Indirectly. The leaderboard highlights active learners (correct answers + learning-in-public links). That visibility helps with networking and recruiter attention, even though the certificate itself is pass/fail.
diff --git a/_questions/machine-learning-zoomcamp/general/108_a84ccd240c_demonstrating-business-impact.md b/_questions/machine-learning-zoomcamp/general/108_a84ccd240c_demonstrating-business-impact.md
@@ -0,0 +1,7 @@
+---
+id: a84ccd240c
+question: How can I show business impact from ML projects? Does the course teach this?
+sort_order: 108
+---
+
+The course focuses on building and deploying real ML services, which gives you practical, portfolio-worthy projects. Business impact depends on the domain context and is best learned on the job. By the end, you’ll have a visible, high-quality portfolio you can discuss with employers. For each project, consider the problem, approach, measurable outcomes, and a clear narrative to make the impact easy to communicate in interviews.
diff --git a/_questions/machine-learning-zoomcamp/general/109_a6f7295637_strategy-pivot-ml-ai.md b/_questions/machine-learning-zoomcamp/general/109_a6f7295637_strategy-pivot-ml-ai.md
@@ -0,0 +1,9 @@
+---
+id: a6f7295637
+question: What strategy do you recommend for an unemployed SWE pivoting to ML/AI soon?
+sort_order: 109
+---
+
+- Ship projects and learn in public daily (posts, blogs, code).
+- Publish everything on GitHub, write about your approach/results, engage on Slack/LinkedIn, and build a network.
+- Use the course projects and optional activities (Kaggle, articles) as portfolio centerpieces.
diff --git a/_questions/machine-learning-zoomcamp/general/111_1e54b4b3ba_why-learn-traditional-ml.md b/_questions/machine-learning-zoomcamp/general/111_1e54b4b3ba_why-learn-traditional-ml.md
@@ -0,0 +1,9 @@
+---
+id: 1e54b4b3ba
+question: Why learn traditional ML if LLMs can do it for me?
+sort_order: 111
+---
+
+Assistants are great accelerators, but you still need **conceptual understanding** to debug, adapt, and own your systems when AI tools make mistakes or hit limits.
+
+The course teaches you to build, reason about, and deploy ML systems you control.
diff --git a/_questions/machine-learning-zoomcamp/module-1/023_68af9cffed_start-module-1-now.md b/_questions/machine-learning-zoomcamp/module-1/023_68af9cffed_start-module-1-now.md
@@ -0,0 +1,7 @@
+---
+id: 68af9cffed
+question: Can I start Module 1 now and move ahead faster?
+sort_order: 23
+---
+
+Yes. Lessons are already available. You can proceed to later modules without waiting for official start dates.
diff --git a/_questions/machine-learning-zoomcamp/module-10/001_0a166909a0_what-tools-are-recommended-for-setting-up-a-local.md b/_questions/machine-learning-zoomcamp/module-10/001_0a166909a0_what-tools-are-recommended-for-setting-up-a-local.md
@@ -8,11 +8,7 @@ sort_order: 1
 Several tools can help set up a local Kubernetes environment:
 
 - **Kind**: Runs Kubernetes clusters in Docker containers, suitable for testing and development.
-
 - **Minikube**: Runs a single-node Kubernetes cluster on your local machine.
-
 - **K3s**: A lightweight Kubernetes distribution ideal for local development.
-
 - **MicroK8s**: A minimal Kubernetes distribution for local development.
-
 - **Docker Desktop**: Includes a standalone Kubernetes server and client for development.
diff --git a/_questions/machine-learning-zoomcamp/module-10/018_23f5be95ef_python-fastapi-deployment.md b/_questions/machine-learning-zoomcamp/module-10/018_23f5be95ef_python-fastapi-deployment.md
@@ -0,0 +1,7 @@
+---
+id: 23f5be95ef
+question: Which language and framework are used for deployment?
+sort_order: 18
+---
+
+Python with FastAPI (updated from Flask). You’ll also package/serve models, containerize, and deploy to cloud/Kubernetes.
diff --git a/_questions/machine-learning-zoomcamp/module-11/001_da9f03ae67_kserve-module-optional.md b/_questions/machine-learning-zoomcamp/module-11/001_da9f03ae67_kserve-module-optional.md
@@ -0,0 +1,7 @@
+---
+id: da9f03ae67
+question: Is the KServe module required?
+sort_order: 1
+---
+
+No. It’s **optional and outdated**. If time-constrained, prioritize Kubernetes.
diff --git a/_questions/machine-learning-zoomcamp/module-4/001_afd915ae5b_module-4-evaluation-overview.md b/_questions/machine-learning-zoomcamp/module-4/001_afd915ae5b_module-4-evaluation-overview.md
@@ -0,0 +1,13 @@
+---
+id: afd915ae5b
+question: What exactly is taught in the evaluation module?
+sort_order: 1
+---
+
+In Module 4, you’ll learn the core evaluation concepts used in classification tasks. Topics include:
+- Metrics and diagnostics: precision, recall, ROC curves, and precision-recall curves
+- Evaluation mindsets: how to think critically about metrics and validation in ML projects
+- Common pitfalls: data leakage, improper validation, misinterpretation of metrics and curves
+- Practical interpretation: selecting metrics based on context (class balance, costs of errors) and conveying results clearly
+- Real-world applicability: how these concepts guide model comparison, threshold selection, and reporting
+This module is designed to be conceptual and abstract, yet practical for real-world ML work.
diff --git a/_questions/machine-learning-zoomcamp/module-5/062_fca4671a69_pretrained-models-deployment.md b/_questions/machine-learning-zoomcamp/module-5/062_fca4671a69_pretrained-models-deployment.md
@@ -0,0 +1,7 @@
+---
+id: fca4671a69
+question: Can I use pre-trained models and focus on deployment?
+sort_order: 62
+---
+
+Yes, you can use pre-trained models and focus on deployment, but you must include some training element (such as fine-tuning, retraining, or a comparable approach). Without a training element, you’ll lose points in training-related criteria, though you can still pass overall if other parts are strong.
diff --git a/_questions/machine-learning-zoomcamp/module-8/001_1d629fe258_how-to-setup-tensorflow-with-gpu-support-on-ubuntu.md b/_questions/machine-learning-zoomcamp/module-8/001_1d629fe258_how-to-setup-tensorflow-with-gpu-support-on-ubuntu.md
diff --git a/_questions/machine-learning-zoomcamp/module-8/001_89f4d77539_how-to-install-tensorflow-in-ubuntu-wsl2.md b/_questions/machine-learning-zoomcamp/module-8/001_89f4d77539_how-to-install-tensorflow-in-ubuntu-wsl2.md
@@ -12,4 +12,6 @@ I was able to get it working by using the following resources:
 - [Install TensorFlow with pip](https://www.tensorflow.org/install/pip#windows-wsl2)
 - [Start Locally | PyTorch](https://pytorch.org/get-started/locally/)
 
-I included the link to PyTorch so that you can get that one installed and working too while everything is fresh on your mind. Just select your options, and for Computer Platform, I chose CUDA 11.7 and it worked for me.
+I included the link to PyTorch so that you can get that one installed and working too while everything is fresh on your mind. Just select your options, and for Computer Platform, I chose CUDA 11.7 and it worked for me.
+
+If you need GPU support, check this article: [https://knowmledge.com/2023/12/07/ml-zoomcamp-2023-project/](https://knowmledge.com/2023/12/07/ml-zoomcamp-2023-project/)
diff --git a/_questions/machine-learning-zoomcamp/module-9/001_5f69e4c975_why-onnx-runtime-deployment.md b/_questions/machine-learning-zoomcamp/module-9/001_5f69e4c975_why-onnx-runtime-deployment.md
@@ -0,0 +1,7 @@
+---
+id: 5f69e4c975
+question: Why ONNX Runtime for DL deployment?
+sort_order: 1
+---
+
+ONNX Runtime provides a cross-framework inference engine for models exported to the ONNX format. If you train in TensorFlow or PyTorch, you can export your model to ONNX and run it with ONNX Runtime, enabling a common deployment path across different environments. This can simplify Lambda/Kubernetes serving by reducing the need to maintain separate framework-specific runtimes (such as TensorFlow Serving or TorchServe) and by providing optimizations available in ONNX Runtime.
diff --git a/_questions/machine-learning-zoomcamp/projects/001_04b9f695e8_could-i-partner-up-for-the-mid-and-final-projects.md b/_questions/machine-learning-zoomcamp/projects/001_04b9f695e8_could-i-partner-up-for-the-mid-and-final-projects.md
@@ -1,7 +1,7 @@
 ---
 id: 04b9f695e8
 question: Could I partner up for the mid and final projects?
-sort_order: 18
+sort_order: 1
 ---
 
 No, you need to do projects individually, but it’s okay to partner up to discuss weekly lectures or exchange ideas.
diff --git a/_questions/machine-learning-zoomcamp/projects/007_9f4011849d_acceptable-datasets-projects.md b/_questions/machine-learning-zoomcamp/projects/007_9f4011849d_acceptable-datasets-projects.md
@@ -0,0 +1,7 @@
+---
+id: 9f4011849d
+question: What datasets are acceptable for projects?
+sort_order: 7
+---
+
+Choose something non-toy (e.g., at least 100 rows) that lets you demonstrate the full ML pipeline: data preparation → modeling → evaluation → deployment. You may also collect your own data.
diff --git a/notebooks/rag.ipynb b/notebooks/rag.ipynb
