Learn / Video Library
Curated software engineering explainers, quick lessons, and practical guides from Abstract Algorithms.
Fresh lessons, loaded directly from YouTube.
Want to master software engineering? Abstract Algorithms is your ultimate all-in-one learning platform. π οΈ Learn how to master Abstract Algorithms Intro in this quick breakdown. π Read the full guide here: [Insert Guide Link Here] In this Short: - Abstract Algorithms: Your Software Engineering Platform - Master System Design, Architecture, and DSA - Master LLM Engineering, RAG, and AI Agents - Visit abstractalgorithms.dev and start learning today! - Like, Share, and Subscribe! π Like, comment, and subscribe for more software engineering guides! #shorts #softwareengineering #coding #programming #tech
Learn something new
4 views β’ Today
Why watch?A focused visual lesson from our channel.
Every time you store data, you probably use an Array. But how does this fundamental data structure actually work under the hood? π οΈ Learn how Arrays work under the hood in this quick breakdown. π Master more data structures here: https://www.abstractalgorithms.dev/learn In this Short: - How do Arrays actually work under the hood? - Contiguous memory allows instant constant-time O(1) lookup. - Inserting or deleting elements in the middle requires O(N) shifting. - Dynamic arrays resize by allocating a new double-sized block. - Master data structures! Read full guides on abstractalgorithms.dev. π Like, comment, and subscribe for more software engineering guides! #shorts #softwareengineering #coding #programming #tech
New β’ Today
Why did a single architecture take over all of AI? From ChatGPT to Gemini, every state-of-the-art model is powered by the Transformer. π οΈ Learn how to master How Transformer Architecture Works: A Deep Dive in this quick breakdown. π Read the full guide here: [Insert Guide Link Here] In this Short: - Why did a single architecture take over all of AI? - Tokens are mapped to vectors with positional encodings. - Self-Attention resolves relationships like 'it' to 'animal'. - Multi-head attention runs in parallel to extract syntax and meaning. - Want to master modern AI? Full guide in description. Subscribe! π Like, comment, and subscribe for more software engineering guides! #shorts #softwareengineering #coding #programming #tech
1 views β’ Today
Fine-tuning a seventy-billion parameter language model once required a supercomputer and eight high-end GPUs. How do we do it on a fraction of the hardware? π οΈ Learn what LoRA and QLoRA are and how they make LLM fine-tuning accessible in this quick breakdown. π Read the full guide here: https://www.abstractalgorithms.dev/fine-tuning-llms-with-lora-and-qlora In this Short: - How do you fine-tune a 70B model on a fraction of the hardware? - Meet LoRA: Freeze the base model and train tiny adapter matrices. - Bottleneck adapters capture the training signal with 99% fewer parameters. - QLoRA compresses the base to 4-bit NF4 to run on a single GPU. - Want to master LLM engineering? Full guide in description. Subscribe for more deep dives! π Like, comment, and subscribe for more software engineering guides! #shorts #softwareengineering #coding #programming #tech
How do you run a massive 70-billion parameter language model on a single consumer GPU? π οΈ Learn what LLM Quantization is and the differences between GPTQ, AWQ, and NF4 in this quick breakdown. π Read the full guide here: https://www.abstractalgorithms.dev/gptq-vs-awq-vs-nf4-choosing-the-right-llm-quantization-pipeline In this Short: How do you run a massive 70B model on a single GPU? Quantization shrinks 16-bit floating point weights to 4-bit integers. GPTQ minimizes layer errors; AWQ protects the most critical weights. NF4 uses a non-uniform scale optimized for weight distributions. Want to master LLM engineering? Full guide in description. Subscribe for more deep dives! π Like, comment, and subscribe for more software engineering guides! #shorts #softwareengineering #coding #programming #tech
2 views β’ Today
Why do neural networks output raw scores like 4.5 or -1.2 instead of actual probabilities? π οΈ Learn what logits are and how temperature controls their randomness in this quick breakdown. π Read the full guide here: https://www.abstractalgorithms.dev/logits-explained In this Short: Why do neural networks output raw scores like 4.5 or -1.2 instead of actual probabilities? These raw, unnormalized scores are logitsβrepresenting the model's raw confidence. Softmax normalizes these raw logits into clean probabilities that sum to 100%. Dividing logits by temperature controls whether output is creative or precise. Want to master machine learning? Full guide in description. Subscribe for more deep dives! π Like, comment, and subscribe for more software engineering guides! #shorts #softwareengineering #coding #programming #tech
Why do AI models fail at simple tasks like counting letters in a word, or reversing a string? π οΈ Learn how to master Tokenization Explained: How LLMs Understand Text in this quick breakdown. π Read the full guide here: https://www.abstractalgorithms.dev/tokenization-explained-how-llms-understand-text In this Short: Why do AI models fail at simple tasks like counting letters or reversing a word? They don't see words or charactersβthey read chunks called tokens. Byte Pair Encoding builds vocab by merging frequent adjacent characters. This determines your API costs, context windows, and spelling bugs. Want to master LLM engineering? Full guide in description. Subscribe for more deep dives! π Like, comment, and subscribe for more software engineering guides! #shorts #softwareengineering #coding #programming #tech
Are database-first models making your Java code fragile and hard to maintain? π οΈ Learn how to master Domain-Driven Design: Designing Aggregates in this quick breakdown. π Read the full guide here: https://www.abstractalgorithms.dev/domain-driven-design-aggregates-value-objects In this Short: * Are database-first models making your Java code fragile and hard to maintain? * Solve it with rich DDD: Entities, Value Objects, and encapsulated Aggregate Roots. * Aggregate Roots act as gatekeepers to validate business invariants. Enforce logic directly in domain classes to guarantee consistency. Want to master DDD? Full guide in description. Subscribe for more deep dives! π Like, comment, and subscribe for more software engineering guides! #shorts #softwareengineering #coding #programming #tech
7 views β’ Today
Are database-first models making your Java code fragile and hard to maintain? π οΈ Learn how to transition to rich Domain-Driven Design (DDD) by protecting invariants inside your model using Entities, Value Objects, and encapsulated Aggregate Roots. π Read the full guide here: https://www.abstractalgorithms.dev/domain-driven-design-aggregates-value-objects In this Short: The problem with database-first "Anemic Domain Models". Entities (Identifiable) vs. Value Objects (Immutable). How Aggregate Roots act as transactional gatekeepers. Rich DDD encapsulated state transitions in Java. If you want to master clean architecture, system design, and software engineering principles, visit https://www.abstractalgorithms.dev π Like, comment, and subscribe for more software engineering guides! #shorts #java #softwareengineering #cleanarchitecture #domaindrivendesign #systemdesign #codingtips
Want to level up your software engineering skills? π Abstract Algorithms is your all-in-one platform featuring in-depth visual guides and roadmaps on: β’ System Design & Scalability β’ Software Architecture & Patterns β’ Data Structures & Algorithms (DSA) β’ Cutting-edge AI Engineering (LLMs, RAG, Agents) Join thousands of developers leveling up their careers today! π Start learning now: https://www.abstractalgorithms.dev/ If you found this video helpful, don't forget to Like, Share, and Subscribe for more software engineering guides! π #SoftwareEngineering #SystemDesign #AIEngineering #Developer #Coding #CodingShorts #Shorts
12 views β’ Today