Dragon

I make AI agents less stupid | Building in public because bugs need witnesses

[Dragon @Dragoncoderx]

10 GitHub Repositories Every AI Engineer Should Bookmark 📌

If you want to become an AI Engineer, don't just learn how models work.

Learn how to build, deploy, evaluate, and scale AI systems.

Here are 10 GitHub repositories worth keeping bookmarked 👇

1. Hands-On Large Language Models

HandsOnLLM/Hands-On-Large-Language-Models

Learn LLMs by building.

Covers:

• Embeddings
• Transformers
• RAG
• Fine-tuning
• Semantic search
• LLM applications

GitHub: https://t.co/bbDCCOUeyP…

2. LLM Course

mlabonne/llm-course

A roadmap for going from LLM fundamentals to advanced topics.

Explore:

• LLM architectures
• Fine-tuning
• Quantization
• RAG
• Agents
• Deployment

GitHub: https://t.co/rFEn36ie5v…

3. Made With ML

GokuMohandas/Made-With-ML

Great for learning how to take ML projects from experimentation to production.

Covers:

• Data
• Modeling
• Evaluation
• Deployment
• MLOps
• Production ML

GitHub: https://t.co/0eisG1uF9Z…

4. GenAI Agents

NirDiamant/GenAI_Agents

Want to understand how modern AI agents are actually built?

Explore examples of:

• Tool use
• Planning
• RAG agents
• Memory
• Multi-agent systems
• Agent workflows

GitHub: https://t.co/dgzZM1bUDX…

5. vLLM

vllm-project/vllm

Learn how LLM inference works at scale.

Useful for understanding:

• High-throughput inference
• Model serving
• KV cache
• Distributed inference
• Production LLM APIs

GitHub: https://t.co/Yweos5rBhy…

6. llama.cpp

ggml-org/llama.cpp

Want to run LLMs locally?

This is one of the most important projects to explore.

Learn about:

• Local LLM inference
• Quantization
• CPU/GPU inference
• Model optimization
• Running models on consumer hardware

GitHub: https://t.co/zOxyowno2a…

7. Haystack

deepset-ai/haystack

A powerful framework for building production AI applications.

Explore:
• RAG
• Pipelines
• Agents
• Retrieval
• Document processing
• LLM applications

GitHub: https://t.co/q5ujl5WY21…

8. LlamaIndex

run-llama/llama_index

Useful for connecting LLMs with your own data.

Learn about:

• RAG
• Data ingestion
• Vector search
• Agents
• Knowledge bases
• Document retrieval

GitHub: https://t.co/kBNnKmR1wM…

9. LiteLLM

BerriAI/litellm

Building applications that use multiple AI providers?
This project is worth exploring.

It provides a unified interface for working with different LLM APIs and models.

GitHub: https://t.co/ySxwdchwgj

10. Prompt Engineering Guide

dair-ai/Prompt-Engineering-Guide

A huge collection of resources for understanding how to work effectively with LLMs.

Explore:

• Prompting techniques
• RAG
• Agents
• LLM research
• Model capabilities
• Prompt optimization

GitHub: https://t.co/3dgsjqtm8y…

Don't try to learn all 10 at once.

Pick based on your goal:

🐍 LLM fundamentals → Hands-On LLMs
🧠 LLM roadmap → LLM Course
🏗️ Production ML → Made With ML
🤖 AI Agents → GenAI Agents
⚡ LLM inference → vLLM
💻 Local AI → llama.cpp
📚 RAG → LlamaIndex / Haystack
🔌 Multiple LLM APIs → LiteLLM
✍️ Prompting → Prompt Engineering Guide

The goal isn't to bookmark more repositories.
The goal is to build with them.

Pick one.
Build a project.

Then move to the next. 🚀

[Dragon @Dragoncoderx]

And on the observability side, I’d also build in tools like OpenLIT, LangSmith/Langfuse, AWS CloudWatch/X-Ray, or GCP Cloud Logging/Trace from day one. They help explain what happened, but the key is still having policy checks and guardrails before the tool call so risky actions never execute in the first place.

[Dragon @Dragoncoderx]

AI agents are easy to demo. Production is harder.

A production-ready agent needs much more than an LLM connected to a few tools.

It needs a workflow that can plan, act, validate, recover, and escalate safely.

A strong end-to-end flow looks like this:

→ Capture the user request and intent
→ Authenticate the user and run safety checks
→ Decide whether the request is allowed
→ Plan the task and break it into clear steps
→ Retrieve context from memory, RAG, vector databases, or knowledge bases
→ Decide whether a tool is needed
→ Select the right tool and verify permissions
→ Execute the action and observe the result
→ Validate accuracy, safety, and output quality
→ Re-plan or retry when the result fails
→ Escalate risky or ambiguous actions to a human
→ Generate the final response
→ Log feedback, metrics, and outcomes for improvement

The important part is what surrounds the core loop.

Monitoring and tracing make behavior visible.
Audit logs support governance and compliance.
Rate limits control abuse and cost.
Secrets management protects credentials.
Privacy policies define what data can be stored.
Model and prompt guardrails reduce unsafe behavior.
Incident handling helps teams respond when something breaks.

That is the difference between an agent that works in a demo and an agent you can trust in production.

The goal is not maximum autonomy.

It is controlled autonomy with clear permissions, validation, observability, and human oversight.

Which layer do you think teams underestimate most when moving AI agents into production?

[Dragon @Dragoncoderx]

@HoustonIntrove1 Exactly, logs are useful, but by then the action has already happened. The real value is stopping a risky tool call before it runs with the right permissions, checks, and approval gates.

[Dragon @Dragoncoderx]

@YashKumarP93601 Love to connect

[Dragon @Dragoncoderx]

Want to become an AI Engineer but don’t know what to learn first?

Don’t try to learn everything.

Pick a path and go deep. 👇

1️⃣ ML Engineer
2️⃣ Deep Learning Engineer
3️⃣ GenAI Engineer
4️⃣ LLM Engineer
5️⃣ RAG Engineer
6️⃣ AI Agent Engineer
7️⃣ MLOps Engineer
8️⃣ Computer Vision Engineer
9️⃣ NLP Engineer
🔟 AI Data Engineer

For every path:

Learn → Build → Deploy → Iterate

The mistake I see often is collecting tools without building anything.

Pick one direction.
Build real projects.
Learn what you actually need along the way.

Depth beats a giant checklist.

Save this for your AI learning roadmap.

#AI #AIEngineering #MachineLearning #GenAI #LLM #AIAgents #MLOps #RAG

[Dragon @Dragoncoderx]

SONNET 5.5 Is here to destroy 🥶

[Dragon @Dragoncoderx]

Gemini 4 ARGON vs GPT-6 LUNA

[Dragon @Dragoncoderx]

#RAG
COMPLETE RAG SYSTEM
@GoogleAI @Gemini @ChatGPT @rag
@webdevjob @workfromhomejobs
#rag
#ml
#rag
#models
#llm
#vectordb

[Dragon @Dragoncoderx]

#fullstackdevelopment
#webdevelopment
#ai
#ml
#chatgpt
#gemini

[Dragon @Dragoncoderx]

🚀 Game-changer alert! Gemini 4 Argon, the latest frontier AI model, is here!
​🛡️ It’s rolling out to cyber defenders STARTING TODAY, with a wider release coming ASAP.
​💡 Incredible progress and amazing introductory pricing! Check the diagram for details! 👇

[Dragon @Dragoncoderx]

┌────────────────────────────────────────────────────────────────────────┐
│ HUMAN / USER PROMPT │
└──────────────────────────────────┬─────────────────────────────────────┘
│
▼
┌────────────────────────────────────────────────────────────────────────┐
│ REASONING & PLANNING ENGINE │
│ ┌───────────────────────┐ ┌────────────────────────────┐ │
│ │ ReAct Loop │ ◄─────────► │ Plan-And-Execute │ │
│ │ (Thought ──► Act) │ │ (Task Decomposition) │ │
│ └───────────────────────┘ └────────────────────────────┘ │
└──────────────────┬──────���───────────────────────────┬──────────────────┘
│ │
▼ ▼
┌──────────────────────────────────────┐ ┌──────────────────────────────┐
│ MEMORY SYSTEM │ │ TOOL EXECUTION │
│ ┌────────────────────────────────┐ │ │ ┌────────────────────────┐ │
│ │ Short-Term: Context Window │ │ │ │ Function Calling / APIs│ │
│ ├────────────────────────────────┤ │ │ ├─────────���──────────────┤ │
│ │ Long-Term: Vector DB / Graph │ │ │ │ Code Execution Sandbox │ │
│ └────────────────────────────────┘ │ │ └────────────────────────┘ │
└──────────────────┬───────────────────┘ └──────────────┬───────────────┘
│ │
��──────────────────┬──────────────────┘
│
▼
┌────────────────────────────────────────────────────────────────────────┐
│ EVALUATION & REFLECTION │
│ ┌───────────────────────┐ ┌────────────────────────────┐ │
│ │ Self-Correction Loop │ ◄─────────► │ Human-in-the-Loop Gate │ │
│ └───────────────────────┘ └────────────────────────────┘ │
└─────���───────────────────────────────┬──────────────────────────────────┘
│
▼
┌──────────────────────────┐
│ FINAL ANSWER │
└──────────────────────────┘

[Logan Kilpatrick @OfficialLoganK]

Introducing Gemini 4 Argon, our new frontier model, rolling out to cyber defenders starting today, and more widely as soon as possible. I am really excited by the progress we have made here. Argon is priced at $2 in and $10 out during introductory pricing!

[Dragon @Dragoncoderx]

Core Pillars of AI Agent Architecture
Perception & Input Processing
Multimodal ingestion (Text, Audio, Vision, Structured Data)
Prompt parsing & contextual retrieval (RAG optimization)
Reasoning & Planning (The Brain)
ReAct Pattern: Interleaving reasoning (thought) and action execution.
Planning Strategies: Tree-of-Thoughts (ToT), Chain-of-Thought (CoT), and reflection/self-correction loops.
Task Decomposition: Breaking complex user goals into sequential sub-tasks.
Memory Systems
Short-term: Active context window management & scratchpads.
Long-term: Episodic and semantic storage using vector databases & graph networks.
Tool Orchestration & Action Execution
Function calling & API integrations.
Dynamic code execution (sandboxed Python interpreters, web scraping, database queries).
Human-in-the-Loop (HITL) approval gates for sensitive actions.
Recommended Learning Path
Level 1: Fundamentals — Master function calling with raw LLM APIs (OpenAI / Anthropic / Google Gemini API).
Level 2: Framework Mastery — Build with Agentic frameworks like LangGraph, CrewAI, or AutoGen to handle stateful multi-agent workflows.
Level 3: Production Engineering — Implement agent tracing/evaluations (LangSmith, Phoenix), error handling, latency optimization, and memory management.

[Dragon @Dragoncoderx]

In Retrieval-Augmented Generation (RAG) and Agentic AI architectures, Indexing and Vector Agents are foundational components responsible for structural knowledge management, fast semantic search, and persistent memory.

[Dragon @Dragoncoderx]

Workflow of RAG

[Dragon @Dragoncoderx]

Rag vs Agentic Rag

[Dragon @Dragoncoderx]

Ai agents

[Dragon @Dragoncoderx]

AI Agents with MCP (Model Context Protocol)

Connecting agents to external tools, databases, APIs, files, and applications.
Very useful for practical agent projects.