Skip to content

ai_course_outline.md

Latest commit

 

History

History
47 lines (42 loc) · 3.5 KB

ai_course_outline.md

File metadata and controls

47 lines (42 loc) · 3.5 KB
Raw

Comprehensive Course on AI and Practical Cyber Applications

Module 1: Introduction to AI and Large Language Models (LLMs)

  • What is Artificial Intelligence? Overview of AI, Machine Learning (ML), and Deep Learning (DL).
  • Introduction to Generative AI & LLMs: What are Large Language Models? How do they process text? (Tokens, Context Windows, Transformers).
  • Prompt Engineering Basics: Best practices for communicating effectively with AI models to get accurate results.

Module 2: The Evolving Landscape of AI Models and Companies

  • Types of Models:
    • Foundational vs. Fine-tuned Models.
    • Open-source (Llama, Mistral) vs. Proprietary (GPT-4, Claude) Models.
  • Major AI Companies and their Offerings:
    • OpenAI: GPT architecture (ChatGPT, GPT-4, GPT-4o).
    • Google: Gemini ecosystem (Nano, Flash, Pro, Advanced).
    • Anthropic: Claude 3 family (focus on safety and large context).
    • Meta: Llama open-source models driving community innovation.
  • Spotlight on Sarvam AI:
    • What is Sarvam AI? India's prominent full-stack AI startup building foundational models.
    • Focus on Indic languages (e.g., OpenHathi), voice-first AI capabilities, and GenAI applications tailored for the Indian demographic.

Module 3: Leveraging Specialized AI Tools for Productivity and OSINT

  • Google NotebookLM for Deep Research:
    • How to turn vast amounts of unstructured data into a grounded knowledge base.
    • Uploading and extracting insights from complex PDFs, cyber reports, and manuals.
    • Generating "Audio Overviews" turning your documents into engaging, podcast-style summaries.
  • Creating Custom Workflows with Gemini Gems:
    • What are Gemini Gems? (Creating personalized expert AI assistants).
    • Building a custom prompt-injected Gem for repetitive analysis tasks (e.g., a "Code Reviewer Gem" or "Malware Analyst Gem").
  • Grok (by xAI) for Social Intelligence:
    • Understanding Grok's integration with real-time data from X (Twitter).
    • Using Grok for Open Source Intelligence (OSINT).
    • Analyzing social profiles, tracking live cybersecurity events, and identifying potential social engineering patterns.

Module 4: Applying AI to Cybersecurity - Mobile APK Analysis

  • Introduction to Android Application Security: Why analyze Android Application Packages (APKs)?
  • The Analysis Workflow: Tools needed to reverse engineer (e.g., apktool, jadx).
  • AI-Assisted Static Analysis:
    • Manifest Analysis: Feeding AndroidManifest.xml to an LLM to automatically flag excessive, dangerous, or anomalous permissions.
    • Code Review: Passing obfuscated or complex compiled Java/Kotlin code snippets to AI to describe what the code is attempting to do.
    • Secret Hunting: Asking the LLM to write regex or identify hardcoded API keys, URLs, and passwords within the decompiled source.

Module 5: Exploring Custom Deployed AI Solutions

  • Custom Security AI Interfaces:
    • Understanding the shift towards application-specific AI deployments (Retrieval-Augmented Generation interfaces).
    • Practical Platform Analysis: Interacting with specialized AI deployments.
    • Link & Lab: Explore Thana GPT as an example of a custom-deployed application running on Google Cloud Run.
    • Discussing architecture, prompt sandboxing, and potential use cases for custom AI web tools within internal teams.