FDE Academy
HomeSign in

Labs library

24 modules · runnable examples · practice labs · client role-plays

Go to my learningOpen the course player
Self-paced courseShort lectures with progress, notes, assessments and a certificate.

The programme

Textbook

Total Specialization: 145–165 Hours

Showing all 24 modules: Courses 1 to 4 plus the Capstone Studio. Beginners should start at Course 1 • Module 1 (Python Data Structures & Big-O).

All 24 modules

Loading textbook modules...

12 Progressive DIY Coding Labs

Starts with 4 Foundation Labs (F1–F4) on Python, SQL, Cosine Similarity & Softmax before the 8 Production FDE Labs (Labs 1–8)!

Lab 1

CO1
The original task statement

🧭 Self-Study Step-by-Step Coding Walkthrough (How to Solve This Lab):
💡 Need Help? Unlock Progressive Hints One-by-One:
Try writing the code yourself first! Click "Unlock Next Hint" above if you get stuck.

Your Python Code Editor (Complete the # TODO sections below)

What Is the AI Agent Code Execution Sandbox & Why Do We Use It?

📘 Plain-English Explanation for Self-Learners: When an enterprise user asks an AI Agent a math or data question (like "Which vendor in our 10,000-row ERP file overcharged us on 18% sales tax?"), LLMs are bad at mental arithmetic—so the AI Agent writes a Python script and sends it to this Execution Sandbox to calculate the exact answer.

Why can't we run AI-generated code directly on our main server? Because if a hacker tricks the AI into writing import os; os.system("rm -rf /"), it would destroy the company server! This workbench lets you complete 4 Guided Real-World Missions using virtual enterprise files (erp_invoices.json and hospital_claims.json) mounted safely inside the sandbox.

Select a Guided Sandbox Mission (1–4)

📂 Mounted Enterprise Files Inside Sandbox

Click a file below to inspect the exact JSON data your sandbox script can open with open("filename.json"):


        


        

          
Mission 1

          

            

            

          

          


          
Agent Python Script (Runs Inside Isolated Ephemeral Sandbox Directory)

          
          

            
            
            
          

          


          
Sandbox Telemetry, AST Security Audit & Standard Output

          
Click "▶ Execute Script in Isolated Sandbox" to run the mission code.

        

      

    


    
    

      

        
6 Chapter-Linked Interactive Simulators & Playgrounds

        
Each playground below corresponds to specific course chapters and includes a self-study guide explaining what to change and what to observe.

        

          
          
          
          
          
          
        

      


      
      

        

          

            
Playground 1 (Chapters 1 & 2): Pipeline SLA & Compound Reliability Simulator

            

              
📘 How to Use This Playground:
              1. Set the number of sequential steps in your AI agent (e.g., 5 steps) and single-step accuracy (e.g., 0.90).
              2. Change "Self-Correction Attempts per Stage" from 1 (No retries) to 3 and click Simulate SLA.
              3. Toggle "Parallel Async Retrieval" on/off to see how running Dense Vector + BM25 concurrently saves 280ms to pass the 2,000ms SLA!

            

            

              
              
              
              
            

            

          

          

            
SLA Simulation Telemetry

            
Click "▶ Simulate Pipeline SLA & Reliability"...

          

        

      


      
      

      
      

      
      

      
      

      
      
    


    
    

      

        
Client simulator

        
Take the FDE role at a client: intake, architecture decisions, sprint build, UAT and the steering committee.

        
How the five stages work
          
 You are the FDE assigned to a paying enterprise client. Each engagement follows the exact 5-stage workflow used on real FDE deployments:
Stage 1 — Client Intake: read messy Slack/email messages from the CRO, CISO and legacy IT team, then write your own Business Requirements Document (BRD) and compare it with the Lead FDE's version.
Stage 2 — Architecture Decisions: make and justify Architecture Decision Records (ADRs) that the CISO/CFO must approve.
Stage 3 — Sprint Build: work through 4 Jira tickets and write a working prototype in Python against the client's real (mounted) dataset. Acceptance tests move tickets from TO DO → DONE.
Stage 4 — Client UAT: the client tries to break your prototype. Edit their dataset with new edge cases and re-run your own code.
Stage 5 — Steering Committee: write speaker notes and risks, generate an executive deck filled with live metrics from your prototype (download → print to PDF), then defend it against tough boardroom questions.

        

        
Choose Your Client Engagement