Data, Databases & the Internet

Data is the fuel

Every AI system depends on data. Photos, transaction logs, customer messages, sensor readings, website clicks — the model learns from examples you provide.

Bad data produces bad AI. Garbage in, garbage out. Before chasing fancy algorithms, ask: Do we collect the right information? Is it accurate? Is it labeled?

Types of data

• check_circleStructured — rows and columns in spreadsheets or databases (sales by date, customer IDs).
• check_circleUnstructured — text, images, audio, video, PDFs.
• check_circleSemi-structured — JSON logs, emails with metadata.
• check_circleTime-series — stock prices, temperature readings, server metrics over time.

Databases explained

Databases organize data for fast retrieval. Relational databases (PostgreSQL, MySQL) use tables with relationships — customers linked to orders. NoSQL databases (MongoDB) store flexible documents — useful for apps with changing schemas.

When you build a business app, choosing the right database affects speed, cost, and how easily you can feed data into AI pipelines later.

How the internet works (simplified)

Mobile apps and websites are clients. AI services like ChatGPT expose APIs your software calls with a prompt and receives a completion.

• check_circleClient — your browser or app requesting information.
• check_circleServer — a computer that responds with web pages, API data, or files.
• check_circleHTTP/HTTPS — protocols for web requests. HTTPS encrypts traffic (look for the padlock).
• check_circleDNS — translates drayengroup.com into a server IP address.
• check_circleAPIs — structured endpoints (e.g. GET /customers/123) machines use to exchange data.

Cloud storage & data pipelines

Modern companies store data in the cloud — S3 buckets, data lakes, warehouses (Snowflake, BigQuery). Data pipelines move and transform raw logs into clean tables analysts and models can use.

ETL (Extract, Transform, Load) is the boring but critical work that makes AI possible at scale.

Privacy & security basics

• check_circleEncrypt sensitive data in transit (HTTPS) and at rest.
• check_circleLimit who can access customer records (role-based access).
• check_circleComply with local laws — Rwanda's data protection rules, GDPR for EU customers.
• check_circleNever paste customer PII into public AI tools without a business agreement.