Applied with AI - Sanjay Kumar

📢 **Update on English Audio**

I’ve seen many of your comments asking for **English audio on my older videos and Members-Only content**. Thank you for bringing this up, and I completely understand the request.

Over the last few days, the English audio versions of my recent videos have started reaching a much wider audience, and many new viewers have also joined the channel membership.

So here’s the update:

I’m currently working on adding **English audio to the Members-Only videos**, and you should start seeing those updates over the **next couple of days**.

I’m also looking at gradually adding English audio to some of the **older public videos**, especially the important and frequently watched ones.

Going forward, I’ll make sure new content is accessible in **both Tamil and English**.

Thank you for your patience, your feedback, and for supporting the channel. ❤️🙏

Exciting things are coming! 🚀

Currently English Audio is available on these videos:
Agentic System Design: https://youtu.be/ZIAzZtKWmbI
AI Roadmaps are a trap: https://youtu.be/JN1mxLsVq28
MCP Explained Right: https://youtu.be/7gQntKsRqTM

1 hour ago (edited) | [YT] | 3

Applied with AI - Sanjay Kumar

Your AI agent confidently gives a wrong answer at times. What’s the best engineering approach to make it more reliable?

4 days ago | [YT] | 19

Applied with AI - Sanjay Kumar

Would you like me to build a series on Google Cloud’s AI Agent stack?

The series would cover "How to Build and Deploy AI Agents on Google Cloud".

5 days ago | [YT] | 6

Applied with AI - Sanjay Kumar

Many users ask the same or similar questions in a RAG app. How can you reduce repeated LLM calls and save cost?

1 week ago | [YT] | 7

Applied with AI - Sanjay Kumar

When do you actually NEED a GPU? 🤔

2 weeks ago | [YT] | 21

Applied with AI - Sanjay Kumar

AI Engineering Project 1 Code Walkthrough is now LIVE exclusively for Channel Members!
www.youtube.com/playlist?list...

Watch all members only video from the above link. I will be publishing more.

In the public video, I explained the complete application demo, architecture flow, and the production AI engineering concepts behind the project.

Now, in this exclusive members-only video, we go inside the actual codebase.

You will see how we implemented:

✅ LangGraph pipeline
✅ Structured output
✅ PII redaction
✅ Prompt injection protection
✅ Token and cost tracking
✅ Ticket Classification

This is where the architecture diagrams become working code.

🔒 The full code walkthrough and source code are available for channel members.

Join the channel and start learning production-grade AI engineering through complete real-world projects.
👉 youtube.com/channel/UCAHa1-TnXAJrkDVMHbcP1Sw/join

3 weeks ago | [YT] | 13

Applied with AI - Sanjay Kumar

Dear viewers,

In my next video, I’ll explain the complete system architecture of a real-world agentic application, layer by layer, through a story-driven walkthrough.

The goal is to show how AI actually fits inside enterprise software, not as a standalone demo, but as part of a production system.

Where does AI Engineering end, and where does Software Engineering begin?

Publishing this week. Stay Tuned.

1 month ago | [YT] | 97

Applied with AI - Sanjay Kumar

🚀 Dear Members, New AI Project Dropped exclusively for Channel Members:
A Multi-Agent Customer Support with Async Human-in-the-Loop. (Members, please find the project inside the "5 Real AI Projects" folder in my drive).

This is the 4th project in the series, that covers multi-turn conversations + multi-agent routing + a human approval step that never blocks the system.

The core idea: an AI can gather info and prep a decision, but a human always pulls the trigger on anything risky (like refunds), and the customer never sits there waiting on a frozen process. This exact pattern shows up in fraud review, content moderation, medical triage, and loan approval systems in the real world.

📚 Concepts Covered:
- Session management (in-memory, real projects use a database or redis)
- Conversational memory (in-memory)
- Context compaction (turning long chats into clean structured JSON retaining context)
- Structured output from LLMs (JSON mode via LiteLLM)
- Multi-agent routing as a state graph
- Asynchronous human-in-the-loop design
- Tool use / function calling with least-privilege DB access
- Separate execution graphs for customer turns vs. approver decisions

If you are not a member and need access to all the AI projects I have shared, please use the below link to join.
youtube.com/channel/UCAHa1-TnXAJrkDVMHbcP1Sw/join

1 month ago (edited) | [YT] | 14

Applied with AI - Sanjay Kumar

Dear Channel Members,
I have added a new project named "RAG_Semantic_Cache_Symposium_Bot" under the "GenAI and RAG" folder in my Google Drive.

Please download the project and practice semantic caching, which is one of the most important concept when you are creating real world RAG apps. Detailed documentation included in the zip file.
Check comments for the link to video explanation of the concept.

If you are not a channel member yet, use this link to join as a member and access all my AI projects shared via members only post.
youtube.com/channel/UCAHa1-TnXAJrkDVMHbcP1Sw/join

2 months ago | [YT] | 74