MLE-Flashcards: 250+ PowerPoint Flashcards for Machine Learning Interview Prep

What MLE-Flashcards Contains and Who Should Use It

MLE-Flashcards is a set of over 250 flashcards assembled by one author over years of ML research, coursework, and independent study. The README describes them as a study and interview preparation reference for people who already have a solid foundation in machine learning. They assume familiarity with technical terminology and use it throughout without introductory definitions.

The intended audience has two segments. Engineers and researchers who already know the material and want to review it systematically before an interview or after a period away from a topic will find the cards useful as-is. Those newer to ML can use the collection to understand the scope of the field and identify gaps, but the README suggests supplementing with introductory materials.

The author notes explicitly that some topics are covered more comprehensively than others, and that the field changes quickly, meaning some content may be out of date. There may also be errors. The collection is a personal reference shared publicly, not a maintained curriculum.

Topics and Structure of the Six PowerPoint Files

The collection distributes across six PowerPoint files at the repository root. Each file covers a distinct domain.

"1 Computer Science.pptx" covers foundational computer science topics including data structures, algorithms, and systems concepts relevant to ML engineers.

"2 Machine Learning General.pptx" covers classical ML: regression, classification, tree-based methods, ensemble methods, regularisation, and evaluation metrics.

"3 Fundamentals for Computer Vision and Deep Learning.pptx" covers neural network architectures, training dynamics, optimisation, convolutional networks, and the fundamentals needed for vision work.

"4 Selected Topics in Computer Vision and Deep Learning.pptx" covers advanced vision topics including 2D and 3D perception, detection, segmentation, and self-supervised learning approaches.

"5 Large Language Models and Related.pptx" covers transformer architectures, pre-training and fine-tuning approaches, instruction tuning, RLHF, and topics in NLP more broadly. The May 2025 update added topics in RL, NeRFs, Gaussian splatting, generative models, LLMs, and VLMs to this and adjacent files.

"ML Flashcards Combined.pptx" is a single file containing all of the above, useful for full-collection review or printing.

How to Get and Use the Collection

The collection is available by cloning or downloading the repository from GitHub. The repository root contains the six PowerPoint files directly, with no build step or installation required. Cloning gives you all files at once:

git clone https://github.com/b7leung/MLE-Flashcards.git

Each file opens in PowerPoint, LibreOffice Impress, or any compatible presentation application. The author switched from an earlier format to PowerPoint in May 2025 specifically for better equation rendering, which affects legibility for mathematical notation in deep learning topics.

The README links to several supplementary resources for those who want more depth: Stanford cs231n, the GenAI Handbook at genai-handbook.github.io, Andrej Karpathy's video lecture series, the Full Stack Deep Learning course, and Chip Huyen's ML Interviews Book at huyenchip.com/ml-interviews-book. These are recommended by the author for filling gaps that the flashcards point to rather than explain.

What the May 2025 Update Added

The repository has two documented update milestones. The initial set of flashcards was released in July 2022. The May 2025 update added topics in reinforcement learning, NeRFs (Neural Radiance Fields), Gaussian splatting, generative models broadly, large language models, and vision-language models.

The NeRF and Gaussian splatting additions reflect the rapid adoption of these techniques in 3D vision between 2022 and 2025. The LLM and VLM additions bring the collection into alignment with the state of the field as of early 2025, covering topics that did not exist in the original 2022 release.

The format change from whatever was used in 2022 to PowerPoint was driven by equation editing quality. Mathematical notation in ML is dense, and the earlier format apparently could not render it as clearly.

The README does not describe a roadmap for future updates or a release cadence, and the repository has no GitHub releases. The last push was on 2026-04-30.

Limitations: No Spaced Repetition, No Interactivity

The collection is static PowerPoint files with no interactivity. It provides no spaced repetition scheduling, no active recall mechanism that hides answers, no quiz mode, and no progress tracking. Learners who want those features need to import the content into a dedicated flashcard platform such as Anki.

The author acknowledges two further limits directly in the README: some topics are covered more comprehensively than others, and there may be errors. This is honest but also means a reader who encounters a card on a topic where they have no prior knowledge cannot verify correctness easily.

The collection is a snapshot of one author's understanding as of May 2025. Topics in LLM architectures and training approaches evolve quickly enough that specific architectural details may be superseded within a year or two of the update.

A well-known alternative with a different approach is the community-maintained Anki deck distributed through AnkiWeb. Those decks vary in quality but offer spaced repetition natively and are maintained by communities of contributors. The tradeoff is that community Anki decks for ML are fragmented across many separate packs, whereas MLE-Flashcards covers the full ML breadth in a single cohesive set from one author's perspective.

Licence, Updates, and Maintenance

MLE-Flashcards is released under the GPL-3.0 licence. GPL-3.0 is a copyleft licence: any modified version distributed to others must also be released under GPL-3.0 with source available. For a collection of presentation slides, this means anyone who modifies the PowerPoint files and distributes them must release their modifications under the same licence.

For individual use, review, and study, the licence imposes no restrictions. The concern applies only when a modified version is distributed externally, for example as part of a commercial course or a paid study package.

The repository has no contributing guidelines for external submissions to the flashcard content itself. Corrections or additions would need to go through GitHub issues or pull requests, and there is no documented process for how the author evaluates or incorporates them.

The last push was on 2026-04-30. The repository has no GitHub releases and no stated release cadence. Updates appear to be driven by the author's own study needs rather than by a maintenance schedule. The collection covers a wide breadth of topics but depth varies by subject: classical ML topics have been in the collection since 2022 and are more developed, while newer additions like Gaussian splatting and VLMs were added in May 2025 and may be thinner in coverage.

Editorial conclusion

MLE-Flashcards is a practical reference for experienced ML practitioners who want a single collection covering the breadth of modern deep learning topics from classical ML through LLMs and VLMs. The PowerPoint format makes it easy to review on any device and to print selected slides, though it provides no spaced repetition, no quiz mode, and no active recall mechanism that dedicated flashcard platforms offer. Newer entrants to ML will find the technical density challenging without supplementary learning materials. The collection was last updated in May 2025, and topics like fast-moving LLM architectures may not reflect the current state of the field. The licence is GPL-3.0, which has implications for redistribution in commercial products.

Frequently asked questions

Official sources

1. b7leung/MLE-Flashcards on GitHub
2. Issues
3. License: GPL-3.0
4. README