Best of LLM — July 2025

1
Article
LangChain·42w
How to Build an Agent
A comprehensive framework for building AI agents from concept to production, covering six key steps: defining realistic tasks with concrete examples, creating standard operating procedures, building an MVP with focused prompts, connecting to real data sources, testing and iteration, and deployment with continuous refinement. The guide emphasizes starting small with well-scoped problems, focusing on core LLM reasoning tasks first, and treating deployment as the beginning of iteration rather than the end of development.
163
2
Article
DigitalOcean Community·42w
LangChain Explained: The Ultimate Framework for Building LLM Applications
LangChain is an open-source Python framework that simplifies building LLM applications by providing standard interfaces for chat models, embeddings, and vector stores. It offers key components like chains for sequential operations, agents for autonomous decision-making, memory for conversation context, tools for external integrations, and vector stores for retrieval-augmented generation. The framework abstracts away complexity when connecting LLMs to external data sources and APIs, making it easier to build chatbots, question-answering systems, and other AI applications without reinventing common functionality.
142
3
Article
ByteByteGo·39w
How Cursor Serves Billions of AI Code Completions Every Day
Cursor is an AI-powered code editor built on VS Code that serves billions of AI completions daily. It features real-time code autocomplete, AI chat assistance, inline editing, and background agents that work in the cloud. The system handles over 1 million queries per second using a distributed infrastructure across AWS, Azure, and GCP, with privacy-focused codebase indexing through vector embeddings. Cursor integrates multiple AI models including GPT-4, Claude, and custom fine-tuned models to provide intelligent coding assistance while maintaining security through encryption and ephemeral data handling.
135
2
4
Article
Javarevisited·40w
10 AI Frameworks and Libraries Every Developer Should Learn in 2025
A comprehensive guide covering 10 essential AI frameworks and libraries for developers in 2025, including LangChain for building LLM applications, vector databases like Pinecone and Weaviate for semantic search, multi-agent systems with CrewAI, fine-tuning techniques like LoRA, and automation tools like N8N. Each framework includes practical use cases and recommended learning resources to help developers build production-ready AI applications.
126
2
5
Article
Javarevisited·41w
Top 5 Books to Learn Prompt Engineering in 2025
A curated list of five essential books for learning prompt engineering in 2025, covering topics from foundational principles to advanced applications. The selection includes practical guides for developers building LLM applications, comprehensive resources on AI engineering infrastructure, specialized books for educational applications, and career-focused materials. Each book targets different audiences from beginners to experienced practitioners, with emphasis on real-world implementation, ethical considerations, and industry best practices.
113
3
6
Article
Javarevisited·41w
Top 5 Books to Learn LLMs (Large Language Models) in Depth
A curated list of five essential books for learning Large Language Models in depth, covering everything from basic engineering concepts to production deployment. The recommendations include practical guides for building LLM applications, training models from scratch, and deploying them at scale. Each book targets different aspects of LLM development, from foundational architecture and prompt engineering to production monitoring and evaluation strategies.
109
3
7
Article
Daily Dose of Data Science | Avi Chawla | Substack·40w
4 Stages of Training LLMs from Scratch
Training large language models from scratch involves four key stages: pre-training on massive text corpora to learn language basics, instruction fine-tuning to make models conversational and follow commands, preference fine-tuning using human feedback (RLHF) to align with human preferences, and reasoning fine-tuning for mathematical and logical tasks using correctness as a reward signal. Each stage builds upon the previous one to create increasingly capable and aligned AI systems.
104
2
8
Article
freeCodeCamp·41w
How AI Agents Remember Things: The Role of Vector Stores in LLM Memory
Large language models don't have inherent memory, but vector stores enable AI agents to simulate memory by converting text into numerical embeddings and storing them in specialized databases. When users interact with AI, the system searches for semantically similar stored vectors to retrieve relevant past information. Popular vector databases include FAISS for local deployments and Pinecone for cloud-based solutions. This approach, called retrieval-augmented generation (RAG), allows AI to appear contextually aware despite technical limitations around similarity-based matching and static embeddings.
98
9
Article
Medium·41w
How to get Kimi-K2 Free API?
Moonshot AI released Kimi K2, a 1 trillion parameter open source model that outperforms Claude 4 Sonnet, GPT 4.1, and DeepSeek V3. While the model requires significant GPU resources to run locally, developers can access it for free through OpenRouter's unified API platform. The guide provides step-by-step instructions to obtain a free API key and includes sample Python code for making requests to the Kimi K2 model through OpenRouter's endpoint.
97
7
10
Article
Javarevisited·40w
Top 7 Project-Based Udemy Courses for AI Engineers in 2025
A curated list of 7 project-based Udemy courses for AI engineers in 2025, focusing on hands-on learning through building real-world applications. The courses cover agentic AI systems, LLM engineering, generative AI with Gemini Pro, automation with n8n, and MLOps deployment. Each course emphasizes practical project development over theoretical learning, helping students build portfolios with technologies like LangChain, OpenAI APIs, CrewAI, and vector databases. The guide includes student enrollment numbers, project counts, and target audience recommendations for each course.
93
11
Article
Tinybird·40w
Why LLMs struggle with analytics
LLMs face significant challenges when working with analytical data, struggling with tabular data interpretation, SQL generation accuracy, and complex database schemas. The key to successful agentic analytics lies in providing comprehensive context through detailed documentation, semantic models, and sample data rather than expecting perfect SQL generation. Building query validation loops with error feedback, using LLM-as-a-judge evaluators, and focusing on business understanding over technical perfection enables more reliable analytical insights.
90
5
12
Article
Machine Learning Mastery·43w
5 Advanced RAG Architectures Beyond Traditional Methods
Five advanced RAG architectures that go beyond traditional retrieval-generation pipelines: Dual-Encoder Multi-Hop Retrieval breaks down complex queries into layered searches; Context-Aware Feedback Loops enable iterative self-improvement through confidence evaluation; Modular Memory-Augmented RAG maintains persistent, contextual memory across sessions; Agentic RAG integrates tool usage for active reasoning and real-time data processing; and Graph-Structured Context Retrieval uses knowledge graphs to find interconnected information rather than simple similarity matches.
78
13
Article
Zed·43w
Why I'm Dialing Back My LLM Usage — Zed's Blog
A seasoned software engineer with 15 years of experience shares his honest reflection on using LLMs in production code. After initially embracing AI tools with enthusiasm, he encountered significant challenges including poor code quality, cascading bugs, and productivity illusions. He advocates for a more measured approach where developers maintain architectural control and use LLMs only for small, well-scoped tasks like refactoring rather than large features. The key insight is treating LLMs as assistants rather than replacements, emphasizing that senior developers should trust their skills and use AI to leverage existing knowledge rather than offload critical decision-making.
77
5
14
Article
Daily Dose of Data Science | Avi Chawla | Substack·42w
MCP Integration with 4 Popular Agentic Frameworks
Part 8 of an MCP crash course demonstrates how to integrate Model Context Protocol with four popular agentic frameworks: LangGraph, CrewAI, LlamaIndex, and PydanticAI. The tutorial provides step-by-step practical walkthroughs for connecting MCP to each framework, along with detailed implementations. This builds on previous parts covering MCP fundamentals, custom client development, tools/resources/prompts, sampling integration, and security considerations including testing and sandboxing.
67
15
Article
Medium·43w
Why PDF Extraction Still Feels LikeHack
PDF extraction remains challenging because the format was designed for print fidelity, not machine readability. Created in 1991 to solve cross-platform document consistency, PDFs treat content as positioned text boxes rather than structured data. Modern AI tools now require complex multi-layer processing (layout analysis, OCR, vision models) to extract meaningful information from PDFs. While Tagged PDF and other standards attempt to add structure, adoption remains limited. The solution involves choosing semantic formats for new content and supporting open standards that preserve both visual fidelity and machine readability.
65
4
16
Article
Javarevisited·43w
Top 5 Educative Courses to Learn AI and LLM Engineering in 2025
A curated list of 5 interactive courses from Educative.io for learning AI and LLM engineering in 2025. The courses cover becoming an LLM engineer, AI for product managers, generative AI essentials, GitHub Copilot mastery, and Cursor AI editor usage. Each course targets different skill levels and roles, from beginners to experienced developers, with hands-on projects and practical implementations. The article also highlights Educative's project-based learning approach and current discount offers.
65
3
17
Article
Getting started with AI·39w
Self-Hosted Private LLM using Ollama and Open WebUI
Learn how to set up a private, self-hosted ChatGPT alternative using Ollama and Open WebUI with Docker. This tutorial covers creating a local AI chatbot that runs entirely on your machine, protecting your data privacy while providing a familiar web interface. The guide includes Docker Compose configuration, model selection and installation, and step-by-step setup instructions for a complete offline AI solution.
64
1
18
Article
Daily Dose of Data Science | Avi Chawla | Substack·40w
What is Context Engineering?
Context engineering is emerging as a critical skill for AI engineers, focusing on systematically orchestrating context rather than just clever prompting. Unlike traditional prompt engineering that relies on 'magic words', context engineering creates dynamic systems that provide the right information, tools, and format to LLMs. The approach addresses the real bottleneck in AI applications - not model capability, but setting up proper information architecture. Key components include dynamic information flow, smart tool access, memory management (both short-term and long-term), and format optimization. As AI models improve, context quality becomes the limiting factor for application success.
62
19
Article
tonsky.me·41w
Gaslight-driven development
Large Language Models are increasingly influencing API design decisions by consistently suggesting certain patterns and method names, forcing developers to adapt their APIs to match AI expectations. Companies like Soundslice and Instant have added features or modified their APIs because LLMs kept referencing non-existent functionality or preferred alternative naming conventions. While this creates a feedback loop where AI shapes the tools it uses, it may also push developers toward more conventional, predictable API designs rather than innovative approaches.
58
10
20
Article
Lobsters·41w
How I keep up with AI progress (and why you must too)
A comprehensive guide to staying informed about AI developments through curated sources and trusted experts. The author provides a structured approach to consuming AI information, starting with foundational sources like Simon Willison's blog and Andrej Karpathy's content, then expanding to official announcements from AI labs, high-signal practitioners in AI engineering, and specialized communities. The guide emphasizes staying close to primary sources, following trustworthy individuals, and building a balanced information diet to avoid both AI hype and dismissal.
46
6
21
Article
Daily Dose of Data Science | Avi Chawla | Substack·41w
Prompting vs. RAG vs. Finetuning
A decision framework for choosing between prompt engineering, RAG, and fine-tuning when building LLM applications. The choice depends on two key factors: the amount of external knowledge required and the level of model adaptation needed. RAG works best for custom knowledge bases without behavior changes, fine-tuning modifies model structure and behavior, prompt engineering suffices for basic adjustments, and hybrid approaches combine RAG with fine-tuning for complex requirements.
44
22
Article
Medium·41w
SmolLM3 : The best small LLM for everything
SmolLM3 is a 3-billion parameter language model from Hugging Face that outperforms larger models through extensive training on 11.2 trillion tokens. Key features include extended thinking mode for step-by-step reasoning, native 64k token context length (extendable to 128k), multilingual support for six languages, and built-in tool calling capabilities. The model excels in benchmarks for math, reasoning, and programming tasks while being deployable on edge devices and single-GPU setups through various frameworks like transformers, vLLM, and llama.cpp.
43
1
23
Article
Daily Dose of Data Science | Avi Chawla | Substack·43w
6 No-code LLM, Agents, and RAG Builder Tools for AI Engineers
Six open-source no-code tools enable AI engineers to build LLM applications, agents, and RAG systems without extensive programming. Featured tools include RAGFlow for document understanding, Langflow for visual agent building, LLaMA-Factory for model fine-tuning, Transformer Lab for local LLM experimentation, xpander for agent backends, and AutoAgent for natural language agent creation. These platforms collectively have over 200k GitHub stars and support various AI development workflows from training to deployment.
40
24
Article
InfoWorld·42w
What you absolutely cannot vibe code right now
Large language models excel at generating repetitive, well-understood code like CRUD applications and web development, but struggle significantly with algorithmic problems and novel implementations. Through practical experience porting a patch system from Python to TypeScript, the author demonstrates that LLMs fail at medium to hard difficulty problems where they cannot rely on well-known templates. While LLMs are valuable tools for routine development tasks, they require human oversight and cannot autonomously handle complex algorithmic design or domains with limited training examples.
38
6
25
Article
Daily Dose of Data Science | Avi Chawla | Substack·40w
Connect Any LLM to Any MCP server
mcp-use is an open-source library that enables developers to connect any LLM to any MCP (Model Context Protocol) server in just 3 lines of code. Unlike being limited to Claude or Cursor, this tool allows building custom MCP agents with local LLMs like Ollama, supports multiple simultaneous MCP server connections, provides sandboxed execution, and includes debugging capabilities for 100% local MCP client development.
36

See all LLM archives