https://www.facebook.com/1260676397123144/

Obianuju Tech, Aba (2026)

16/10/2026

DAY 45 — MODEL ROUTING: WHY ONE AI MODEL SHOULDN'T DO EVERYTHING

📖 WHAT YOU NEED TO KNOW
Yesterday, we explored Hybrid Search—how AI combines different search methods to retrieve more relevant information.

But here's another interesting question:
Why should an AI system use the same model for every task?
Think about it.

Would you hire an accountant to design your company logo?

Or ask a graphic designer to investigate a cybersecurity incident?
Probably not!

Different tasks require different skills.
The same principle applies to AI.
That's where Model Routing comes in.

💡 WHAT IS MODEL ROUTING?
Model Routing is the process of selecting the most suitable AI model to handle a particular task.
Instead of sending every request to one powerful and expensive model, the system chooses based on the task's requirements.
For example:
⚡ Simple questions → A smaller, faster model.
🧠 Complex reasoning → A model with stronger reasoning capabilities.
💻 Code generation → A model suited to programming tasks.
🖼️ Image analysis → A vision-capable model.
🔐 Sensitive tasks → A model approved for the required security and privacy environment.

The goal isn't always to use the biggest model. It's to use the right model for the job.

⚙️ PRACTICAL DEVSECOPS EXAMPLE
Imagine building an AI Assistant for your DevOps team.
Throughout the day, engineers ask it to:
📝 Summarize deployment reports.
💻 Review application code.
🔐 Investigate security vulnerabilities.
☁️ Analyze cloud infrastructure configurations.
📊 Generate daily operational summaries.
Should every task use the same AI model?
Not necessarily!

With Model Routing:
✅ Simple summaries can use a lightweight model.
✅ Complex code reviews can use a stronger coding or reasoning model.
✅ Security investigations can use a model evaluated for security analysis.

The routing system directs each request to an appropriate model based on its requirements.
This can improve efficiency without unnecessarily increasing costs.

🔄 HOW MODEL ROUTING WORKS
Here's a simplified workflow:
1️⃣ User Request: Someone asks the AI to complete a task.
2️⃣ Task Analysis: The system identifies what the task requires.
3️⃣ Model Selection: The router selects a suitable model.
4️⃣ Processing: The chosen model handles the request.
5️⃣ Response: The result is returned to the user.

For example:
User → Model Router → Selected AI Model → Response

Some routing systems can also switch to another approved model if the preferred one becomes unavailable.

WHY THIS MATTERS
Model Routing helps organizations:
✅ Reduce unnecessary AI costs.
✅ Improve response speed.
✅ Match model capabilities to tasks.
✅ Support multiple AI models.
✅ Improve availability through fallback options.
✅ Apply model-specific privacy and security requirements.

But remember:
A smaller model isn't automatically the best choice for every simple-looking task.

Good routing requires testing and evaluation to ensure the selected model delivers acceptable results.

🧠 THE BIG PICTURE
Let's connect our recent lessons:
📚 RAG: Helps AI retrieve external knowledge.
🧠 Vector Embeddings: Represent information based on meaning.
🗄️ Vector Databases: Store and search embeddings.
🔗 Knowledge Graphs: Connect related information.
🔍 Hybrid Search: Combines search methods for better retrieval.
🤖 Model Routing: Selects an appropriate AI model for each task.

Together, these technologies help us design more efficient and capable AI systems.
Building intelligent AI isn't just about having powerful models. It's about knowing when and how to use them.

🎯 TODAY'S EXERCISE
Imagine you're building an AI Assistant for your organization.
It handles two requests:
Task A: Summarize a short meeting note.
Task B: Investigate a critical Kubernetes security incident.

Would you route both tasks to the same AI model?
Why or why not?

Share your thoughts in the comments!

Tomorrow: Day 46 — Model Serving: How AI Models Handle Requests Behind the Scenes.

15/10/2026

DAY 44 — HYBRID SEARCH: WHY ONE SEARCH METHOD ISN'T ENOUGH

📖 WHAT YOU NEED TO KNOW
Yesterday, we explored Knowledge Graphs—how AI connects information and understands relationships between people, applications, systems and data.

But here's another interesting question:
What happens when AI needs to find information using different search methods?

Sometimes we know the exact information we're looking for.

Other times, we only know what it means.
That's where Hybrid Search comes in.

Hybrid Search combines different search techniques to retrieve more relevant information.

💡 THINK OF A LIBRARY
Imagine visiting a library.
Sometimes you know the exact title of a book.
Other times, you only remember what the book is about.

Would you search the same way in both situations?
Probably not!

AI faces the same challenge.

Keyword Search finds information using specific words or phrases.

Vector Search finds information based on meaning and similarity.

Hybrid Search combines their strengths.

⚙️ PRACTICAL DEVOPS EXAMPLE
Imagine building an AI Assistant to troubleshoot Kubernetes issues.
You ask:
"Why is my Pod showing CrashLoopBackOff?"
The AI could use:
🔎 Keyword Search → Find documentation containing the exact error CrashLoopBackOff.

🧠 Vector Search → Find related information about container crashes, startup failures and incorrect configurations.

📚 Hybrid Search → Combine both sets of results to retrieve more useful troubleshooting information.
Instead of relying on one search method, the AI gets a broader collection of relevant evidence.

🔄 HOW HYBRID SEARCH WORKS
Here's a simplified workflow:
❓ A user asks a question.
🔎 Keyword search looks for matching terms.
🧠 Vector search looks for similar meanings.
🔗 The system combines and ranks the results.
🤖 Relevant information is provided to the AI to generate an answer.
The goal is to retrieve the most useful information, not simply the most documents.

🔐 WHY THIS MATTERS
Hybrid Search can help improve:
✅ Retrieval accuracy
✅ Search relevance
✅ Handling of exact technical terms
✅ Understanding of related concepts
✅ RAG application performance

For example, searching for an exact error code requires a different approach from searching for general troubleshooting advice.
Hybrid Search helps handle both.
However, it doesn't automatically guarantee correct answers. Retrieved information still needs to be relevant, accurate and properly authorized.

🚀 THE BIG PICTURE
Let's connect our recent lessons:
📚 Day 40 — RAG: Retrieves external information to support AI responses.
🧠 Day 41 — Vector Embeddings: Represent information numerically to capture similarity.
🗄️ Day 42 — Vector Databases: Store and search embeddings.
🔗 Day 43 — Knowledge Graphs: Connect information through relationships.
🔍 Day 44 — Hybrid Search: Combines search methods to retrieve more relevant information.

Together, these technologies help us build AI systems that can find and use information more effectively.

The smartest search system isn't necessarily the one using the most advanced technique. It's the one using the right combination of techniques for the task.

🎯 TODAY'S EXERCISE
Imagine you're building an AI Assistant for your DevOps documentation.
A user asks:
"Find the solution to error code 503 in my Kubernetes application."

Would you use keyword search, vector search or both?

Why?

Share your thoughts in the comments!

Tomorrow: Day 45 — Chunking: Why Breaking Documents Into Smaller Pieces Improves RAG.

14/10/2026

5 THINGS SCAMMERS HOPE YOU WON'T CHECK BEFORE CLICKING A LINK!
That message claiming your bank account will be blocked?
That delivery notification you weren't expecting?
Or that familiar voice asking you to open a link urgently?
Not everything that looks legitimate is safe.
Cybercriminals use phishing, impersonation, fake websites, and even AI-generated voices to trick people into sharing sensitive information.
In today's Digital Safety & Scam Awareness episode, we're discussing five warning signs everyone should know.
🔐 Remember: PAUSE → INSPECT → VERIFY
💬 Have you ever received a suspicious link or message? What made you realize something was wrong?

Share your experience, but please don't post suspicious links or personal information in the comments.

14/10/2026

DAY 43 — KNOWLEDGE GRAPHS: HOW AI CONNECTS INFORMATION

📖 WHAT YOU NEED TO KNOW

Yesterday, we explored Vector Databases—how AI systems find relevant information based on meaning rather than exact keywords.

But here's another interesting question:
What happens when AI needs to understand how different pieces of information are connected?

That's where Knowledge Graphs come in.
A Knowledge Graph organizes information by connecting entities (people, places, applications or things) through their relationships.

💡 THINK OF A FAMILY TREE
Imagine your family tree.
It shows more than just names.
It shows relationships:
👨‍👩‍👧 Parents have children.
👫 People have siblings.
👵 Grandparents have grandchildren.

These connections help you understand how everyone is related.
Knowledge Graphs work similarly.
Instead of treating information as isolated facts, they connect related information.

⚙️ PRACTICAL DEVSECOPS EXAMPLE
Imagine an AI Security Agent investigating a vulnerable software library.
The organization has several applications running across different cloud environments.
A Knowledge Graph might represent relationships such as:
📦 Library A → Used by Application B
💻 Application B → Managed by Team C
☁️ Application B → Deployed on Cluster D
🔐 Cluster D → Supports Business Service E

Now imagine a security vulnerability is discovered in Library A.

Instead of searching through unrelated documents, the AI can follow these relationships to identify:
✅ Which applications may be affected.
✅ Which engineering teams are responsible.
✅ Which cloud environments need investigation.
✅ Which business services could be at risk.

That's the power of connecting information.

🔍 KNOWLEDGE GRAPHS VS VECTOR DATABASES
Both technologies help AI work with information, but they serve different purposes.

🗄️ Vector Databases
Help find information with similar meanings.
Example: Finding documentation related to Kubernetes networking problems.

🔗 Knowledge Graphs
Help identify relationships between entities.
Example: Finding which applications depend on a vulnerable Kubernetes component.
Vector Databases help answer: "What's similar?"
Knowledge Graphs help answer: "What's connected?"

🚀 WHY THIS MATTERS
Knowledge Graphs are useful in:
🔐 Cybersecurity investigations
💻 Software dependency analysis
🏢 Enterprise knowledge management
🛒 Recommendation systems
📚 AI-powered learning platforms
🤖 Intelligent AI Agents

They help AI applications navigate complex relationships instead of relying only on isolated information.

And when combined with vector search, they can support more powerful knowledge retrieval systems.

THE BIG PICTURE
Let's connect our recent lessons:
📚 Day 40 — RAG: Retrieves external knowledge to support AI responses.

🧠 Day 41 — Vector Embeddings: Represent information numerically to capture similarities.

🗄️ Day 42 — Vector Databases: Store and search embeddings efficiently.

🔗 Day 43 — Knowledge Graphs: Connect entities and their relationships.

Together, these technologies help AI systems retrieve information and understand how different pieces of knowledge relate.
Finding information is useful. Understanding how that information connects can make it even more valuable.

🎯 TODAY'S EXERCISE
Imagine you're building an AI Assistant to manage your cloud infrastructure.
You have:
• Applications
• Kubernetes clusters
• Databases
• Engineering teams

What relationships would you create between these entities in a Knowledge Graph?

Share your thoughts in the comments!

Tomorrow: Day 44 — GraphRAG: Combining Vector Search and Knowledge Graphs for Smarter AI Retrieval.

13/10/2026

DAY 42 — VECTOR DATABASES: WHERE AI STORES AND FINDS MEANING

📖 WHAT YOU NEED TO KNOW

Yesterday, we explored Vector Embeddings—how AI represents information as numbers to help identify similar meanings.

But here's an interesting question:

After converting thousands or millions of documents into embeddings, where are they stored, and how does AI find the right ones?

That's where Vector Databases come in.

💡 THINK OF A SMART LIBRARY

Imagine walking into a huge library containing millions of books.

Instead of asking for a book by its exact title, you say:

"I want books about securing cloud applications."

A smart librarian finds books covering cloud security, access control, encryption and other related topics.

Even when those books don't contain your exact words!

That's similar to how a vector database works.

It helps applications find information based on mathematical similarity rather than relying only on exact keyword matches.

⚙️ PRACTICAL DEVOPS EXAMPLE

Imagine building an AI Assistant to help troubleshoot Kubernetes problems.

Your documentation contains information about:

📄 CrashLoopBackOff errors 📄 Pod scheduling failures 📄 Container resource limits 📄 Network connectivity 📄 Application configuration

Now, you ask:

"Why does my Kubernetes Pod keep restarting?"

Instead of searching only for those exact words, a vector database can help retrieve documentation related to container crashes, failed health checks and other relevant issues.

The AI can then use that information to provide a more useful troubleshooting response.

🔄 HOW VECTOR DATABASES WORK

Here's a simplified workflow:

* 📄 Documents are converted into vector embeddings.

* 🗄️ The embeddings are stored and indexed.

* ❓ A user's question is converted into an embedding.

* 🔍 The database searches for similar vectors.

* 🤖 Relevant documents are retrieved and provided to the AI.

This is one of the common retrieval approaches used in Retrieval-Augmented Generation (RAG).

🛠️ POPULAR VECTOR DATABASE TECHNOLOGIES

Some examples include:

• Pinecone • Weaviate • Qdrant • Milvus • Chroma • PostgreSQL with pgvector

Interestingly, you don't always need a completely separate database.

For example, PostgreSQL can support vector similarity search through the pgvector extension.

🔐 WHY THIS MATTERS

Vector databases help power:

✅ Semantic search ✅ RAG applications ✅ AI-powered knowledge assistants ✅ Recommendation systems ✅ Similarity-based document retrieval

But remember, vector databases don't replace traditional databases.

Traditional databases are excellent for structured queries and exact records.

Vector databases specialize in similarity searches.

Many modern applications use both.

THE BIG PICTURE

Let's connect our last three lessons:

📚 Day 40 — RAG: Helps AI retrieve external knowledge before answering.

🧠 Day 41 — Vector Embeddings: Represent information numerically so similar meanings can be identified.

🗄️ Day 42 — Vector Databases: Store, index and search embeddings efficiently.

Together, they help AI systems retrieve relevant knowledge at scale.

Embeddings represent meaning. Vector databases help find similar meaning. RAG uses retrieved information to generate grounded answers.

🎯 TODAY'S EXERCISE

Imagine you're building an AI Assistant for your DevOps documentation.

Would you choose a traditional database, a vector database, or both? And why?

Share your thoughts in the comments!

Tomorrow: Day 43 — Chunking: Why Breaking Documents Into Smaller Pieces Improves RAG.

13/10/2026

The Technology Behind Bank Transfers.

💬 Have you ever experienced a failed transfer or delayed payment?

12/10/2026

YOU TRANSFER MONEY IN SECONDS. BUT HAVE YOU EVER WONDERED WHAT HAPPENS BEHIND THE SCENES?

Welcome to our new Monday series on Obianuju Tech! 💙

Every Monday, we'll explore the technology behind things we use in everyday life.

Not just definitions. Not just technical jargon.

Real technology. Real applications. Real-world problems.

Let's begin with something familiar.
Imagine you want to send ₦10,000 to a friend.

You open your banking app, enter the amount, confirm the transaction, and within seconds, you receive a notification.

Transfer successful! ✅

Simple, right?

But behind that little success message, several technologies are working together.

Let's explore five of them.
1️⃣ APIs — The Messengers
When you click Send, your banking app communicates with backend systems through APIs (Application Programming Interfaces).

Think of an API like a waiter in a restaurant.
You place an order, the waiter takes it to the kitchen, and brings back the response.
Your banking app does something similar.
It sends your transaction request to the appropriate backend service and receives a response.

2️⃣ Backend Systems — The Decision Makers
The backend processes your request.
It checks things like:
Is the transaction properly authorized?
Is the account allowed to make this transfer?
Are sufficient funds available?
Does the transaction meet the bank's security requirements?
These checks help determine whether the transfer should proceed.

3️⃣ Databases — The Record Keepers
Banks need reliable records of financial transactions.

Databases help store account information, transaction details, and other important records.

But there is something even more important.
A banking system must maintain accurate financial records even when multiple transactions happen simultaneously.
Imagine two transfers trying to spend the same money!
That's why transaction integrity and consistency matter.

4️⃣ Cloud & Infrastructure — The Foundation
Thousands or millions of customers may be using banking services.
The infrastructure supporting those services needs to handle traffic, maintain availability, and recover from failures.

Now imagine someone trying to transfer money from an account they don't own.
Scary, right?

Financial systems use multiple security controls, including authentication, authorization, encryption, fraud detection, and transaction monitoring.
Security isn't something that should be added after the application is built.
It needs to be considered throughout the system.

So where does DevOps come in?
DevOps practices help engineering teams build, test, deploy, monitor, and maintain applications more reliably.

And DevSecOps brings security considerations into those workflows.
This is one reason I'm fascinated by the engineering behind everyday technology.
A simple button on your screen can depend on many interconnected systems working correctly.

And here's something worth remembering:
The best technology often looks simple to the person using it because of the complex engineering happening behind the scenes.
Next time you make a bank transfer, remember that there's much more happening than just tapping Send.
💬 Let's discuss:
Have you ever experienced a failed bank transfer where your account was debited but the recipient didn't receive the money immediately?

Welcome to Tech Behind Everyday Life!

Follow Obianuju Tech as we continue making technology easier to understand.

12/10/2026

A small configuration change can stop an entire application from working.

In my latest DevSecOps exercise, the database was running, but the backend couldn’t find it.

Think of two people in separate rooms with no connecting door. Both are present, but they cannot communicate.
That was the problem with the containers: the backend and database were connected to different Docker networks.

I checked the logs, inspected their networks, and restored a shared network. The backend could then find the database using its service name, db.

After the fix:
✅ Both services became healthy.
✅ The backend connected to the database.
✅ A test user was registered and saved successfully.
✅ All six automated verification checks passed.

This is a practical skill I can apply when troubleshooting application deployments, investigating service connection failures, and checking whether configuration changes have broken a working system.

My biggest takeaway: understand the error before changing the settings.

12/10/2026

DAY 41 — VECTOR EMBEDDINGS: HOW AI SEARCHES BY MEANING, NOT JUST WORDS

📖 WHAT YOU NEED TO KNOW

Yesterday, we explored Retrieval-Augmented Generation (RAG)—how AI retrieves relevant information from external sources before generating answers.

But here's an interesting question:

How does AI find the right information among thousands of documents?

Does it only search for matching words?

Not necessarily!

That's where Vector Embeddings come in.

💡 THINK OF A LIBRARY

Imagine asking an AI Assistant:

"How do I apply for annual leave?"

But your company's HR handbook uses the phrase:

"Employees are entitled to paid vacation."

The words are different, but the meaning is similar.

A basic keyword search might miss that connection.

However, AI systems using embeddings can recognize that annual leave, vacation, and paid time off are related concepts.

How?

By representing information as numbers!

WHAT ARE VECTOR EMBEDDINGS?

Vector embeddings are numerical representations of information that help AI systems identify patterns and similarities.

Think of them as giving words, sentences, or documents a mathematical address.

Information with similar meanings is often positioned closer together in this mathematical space.

For example:

📄 "How do I secure my application?"

📄 "Best practices for application security."

These sentences use different words but express related ideas.

Embeddings help AI recognize that relationship.

⚙️ PRACTICAL DEVOPS EXAMPLE

Imagine building an AI Assistant for your DevOps documentation.

You ask:

"Why is my Kubernetes application not starting?"

Your documentation might contain information about:

🔍 CrashLoopBackOff errors
📋 Container startup failures
🛠️ Incorrect configurations
📊 Application logs
🔐 Missing environment variables

Even without matching your exact question, vector embeddings can help retrieve relevant troubleshooting documents.

That's semantic search—finding information based on meaning, not just exact keywords.

🔄 HOW EMBEDDINGS WORK WITH RAG

Here's a simplified workflow:

📄 Documents are converted into embeddings.

🗄️ Embeddings are stored in a searchable index, often using a vector database.

❓ The user's question is converted into an embedding.

🔍 The system searches for similar embeddings.

Relevant information is retrieved and provided to the AI to generate an answer.

Embeddings help find relevant information, but the retrieved results still need to be checked for accuracy and relevance.

WHY THIS MATTERS

Vector embeddings help power:

✅ Semantic search
✅ RAG applications
✅ Recommendation systems
✅ Document similarity detection
✅ Knowledge retrieval
✅ AI-powered assistants

They're useful because people don't always use the same words to describe the same thing.

RAG helps AI retrieve external knowledge.

Embeddings help AI find related information based on meaning.

Together, they make AI-powered knowledge systems more useful.

🎯 TODAY'S EXERCISE

Imagine building an AI Assistant for your technical documentation.

A user asks:

"Why does my container keep restarting?"

What related technical terms or errors should a semantic search system consider?

Share your answers in the comments!

Tomorrow: Day 42 — Vector Databases: Where AI Stores and Searches Embeddings.

11/10/2026

DAY 40 — RETRIEVAL-AUGMENTED GENERATION (RAG): WHY AI DOESN'T NEED TO KNOW EVERYTHING

📖 WHAT YOU NEED TO KNOW

Yesterday, we celebrated Day 39 of our 55 Days of AI journey, reflecting on everything we've learned about building intelligent AI systems.

Today, we're exploring another important concept:

Retrieval-Augmented Generation (RAG).

Have you ever asked AI a question about your company's documents, only to receive a generic or incorrect answer?

That's because AI doesn't automatically know everything about your organization.

This is where RAG becomes useful.

💡 THINK OF A LIBRARIAN

Imagine walking into a library and asking for information about a particular subject.

The librarian doesn't need to memorize every book.

Instead, they know how to find the right information and help you understand it.

RAG works in a similar way.

It allows AI to retrieve relevant information from external sources before generating an answer.

⚙️ PRACTICAL EXAMPLE

Imagine asking an AI Assistant:

"What is our company's leave policy?"

❌ Without RAG:

The AI might give a general answer based on its training, which may not match your company's actual policy.

✅ With RAG:

The AI can:

1. 🔍 Search your company's HR documents.
2. 📄 Retrieve the relevant leave policy.
3. 🧠 Use that information as context.
4. 💬 Generate an answer based on the retrieved document.

The AI model hasn't changed.

It simply has access to more relevant information.

🔄 HOW RAG WORKS

Here's the basic workflow:

Question → Search → Retrieve → Add Context → Generate Answer

RAG can retrieve information from:

📁 Company documents
🗄️ Databases
🌐 Knowledge bases
💻 GitHub repositories
☁️ Cloud storage
📚 Technical documentation

This helps AI answer questions using information beyond its original training data.

🔐 WHY THIS MATTERS

RAG can help organizations:

✅ Access updated information without retraining the model.

✅ Generate answers grounded in relevant documents.

✅ Reduce unsupported responses.

✅ Provide references to information sources.

✅ Make internal knowledge more useful.

But remember:

RAG doesn't guarantee accurate answers.

If the retrieved information is outdated, incorrect or irrelevant, the AI can still produce misleading responses.

Access permissions and data security also matter.

🚀 THE BIG PICTURE

We've learned that:

💭 Memory helps AI retain useful information.

🧠 Context Engineering provides the right information for a task.

📚 RAG retrieves relevant knowledge from external sources when needed.

Together, these approaches help us build more useful and reliable AI systems.

AI doesn't need to memorize everything. It needs a reliable way to find and use the right information.

🎯 TODAY'S EXERCISE

Imagine building an AI Assistant for your workplace.

What 3 types of documents would you connect to it using RAG?

And how would you ensure it only accesses information users are authorized to see?

Share your thoughts in the comments!

Tomorrow: Day 41 — Vector Embeddings: How AI Searches by Meaning, Not Just Keywords.

Obianuju Tech

16/10/2026

15/10/2026

14/10/2026

14/10/2026

13/10/2026

13/10/2026

12/10/2026

12/10/2026

12/10/2026

11/10/2026

Address

Website

Alerts

Shortcuts

Share

Category

Listing takedown request