Three-day, hands-on AI workshop · 2–4 November 2026
Website (https://decoding-data-science.github.io/sharjah/)
https://decoding-data-science.github.io/sharjah/prompt-engineering-guide.html#five
https://decoding-data-science.github.io/sharjah/rag_play_lab.html
Individual Survey
https://forms.gle/SLs9ksmtvqWjK6276
Team Capstone Submission
https://forms.gle/9FcgqLh6zFyW4ZgF8
DAY 2 RAG Sample prompt to improve gradio after final step
So please, for the above Gradio— everyone please listen to me, right? For the above Gradio application, please make it more professional. Use the same logic, don't change the logic. Only make sure that you are changing the Gradio component, making it more professional. Use the package Gradio 6.31.0. Make sure whatever settings you are giving, make it for Gradio 3.1.0. What else we can do here? Try to centrally align our title. Customize this for the team. What's your team name? AI Vision. Add relevant logo. Make two-column layout. Add FAQ on the left side. Make it more professional. Add layouts, blocks, borders. And what else? Anything more? Add icons where required. Anything else to make it beautiful? Add background images as relevant. Anything else? How to make it better? The Gradio? Okay, let us try to do this much, and make sure that you are using the same logic and function. Don't change anything.
Day 3 Sample prompt to improve gradio after final step
Can you please help me make the above Gradio application professional? Make sure you are doing all the changes in 6.31.0 Gradio version. Please don't use any features like button features and copy features that are there in the old Gradio. Please make two layouts. Put the logo. Please design it for the team Desert Dragons. Keep a relevant logo, keep a relevant design feature. Please add relevant numbers, dashboards, or important views. Add a status symbol like red, green, yellow, just for us to have a good view. Add FAQ questions. Add borders, layouts. Make it more interesting, professional, and make sure that you don't change any logic. Only use the logic that is there inside the Gradio. Also try to make it something similar to how the ChatGPT interface looks like, with the logo should be on the left side, and the team name Desert Dragon should be shown as well, and add relevant icons where required.
Day 3 Focus on team use case
So now I need to create a focused Gradio application based on the above that you have built, focusing on customer feedback analysis. So have one tab for the agent, whatever you have built, FAQs, which is a little bit focused and specialized on customer feedback data. Second tab, please add some metrics, some KPIs that show about the feedbacks, like what are the top segments. If you can give me some monthly important dashboards and KPIs that are there inside the agent within a new tab. So please add it and convert this specific general agent to the agent that is focused on the above details and also give me some good, important metrics dashboard-related graphs that is very helpful for the person who logs in. use the same gadio vesion just add extra features as discussed above
A practical introduction to data analytics, generative AI, document understanding, and AI-assisted workflows using Google Colab and the Gemini API. The exercises follow a fictional airport operations case study known as “The Friday Problem.”
Workshop safety: All sample data and policies are fictional. Do not upload real passenger, employee, organizational, or otherwise confidential information to AI tools. Store API keys in Google Colab Secrets—not in code or this repository.
| Day | Focus | Outcomes |
|---|---|---|
| Day 1 — Explore & Ask | Data understanding, document understanding, models, system prompts, and context | Explore airport data and ask better grounded questions. |
| Day 2 — Build the Analyst | AI-assisted analysis and practical notebooks | Apply the tools to structured and unstructured workshop materials. |
| Day 3 — Put It Together | Integrated workflow and practice | Combine the skills into an end-to-end use case. |
- Read the participant handbook and workshop handout.
- Open the workshop website for the daily agenda, setup, exercises, and downloads.
- Prepare a Google account and open Google Colab.
- Obtain any required Gemini API access through Google AI Studio and keep credentials in Colab Secrets.
- Work through the Day 1 notebooks in numerical order. Open each
.ipynbin Colab and follow its setup cells. - Use the fictional airport data and document register for exercises.
The synthetic SQLite database is included as eight compressed parts under data/database/. After cloning or downloading the repository, restore the original airport_ops.db by running:
python data/database/restore_database.pyThe script checks the database integrity before writing it. The Excel workbook with 11 worksheets is available directly, without any restoration step. See the database README for details.
sharjah/
├── README.md # Start here
├── .gitignore
├── docs/ # GitHub Pages-ready workshop website
│ ├── index.html
│ ├── setup.html
│ ├── day-1.html
│ ├── day-2.html
│ ├── day-3.html
│ ├── exercises.html
│ ├── downloads.html
│ ├── assets/
│ └── downloads/ # Participant downloads released for Day 1
├── notebooks/ # Day 1 Colab notebooks and Python helpers
│ ├── D1_01_Data_Understanding.ipynb
│ ├── D1_02_Documents_Understanding.ipynb
│ ├── D1_03_Models_System_Prompt_Context.ipynb
│ ├── workshop_helpers.py
│ └── workshop_agent.py
├── data/
│ ├── database/
│ │ ├── airport_ops.db # Rebuild with restore_database.py
│ │ └── Data_Dictionary.md
│ └── documents/ # Fictional policy documents in TXT/PDF
├── slides/ # Day 1 presentations
├── handouts/ # Participant PDFs
├── release_later/ # Day 2/3 material: publish on schedule
├── source/ # Workshop content-generation scripts
└── index_single_file.html # Portable, single-file version of the site
Note: The folder tree describes the full workshop distribution. If some assets are not yet available in GitHub, use the original workshop bundle until they are published.
| Resource | Location |
|---|---|
| Website and day-by-day guidance | docs/ |
| Colab notebooks | notebooks/ |
| Synthetic database and dictionary | data/database/ |
| Fictional policies and document register | data/documents/ |
| Presentation slides | slides/ |
| Participant handbook and handout | handouts/ |
| Day 2 and 3 release materials | release_later/ |
| Site and resource generator scripts | source/ |
The prepared website uses plain HTML, CSS, and JavaScript—no site build is required.
- In GitHub → Settings → Pages, choose Deploy from a branch.
- Choose branch main and folder /docs.
- Save. Once the site is deployed, visit https://decoding-data-science.github.io/sharjah/.
Release policy: The Day 2 and Day 3 participant assets live in release_later/. Copy each day's intended materials into docs/downloads/ only when that day's release is approved. Avoid publishing facilitator solutions or answer keys.
- Use only the synthetic airport database and fictional policy documents provided.
- Confirm the effective/current document version; some included policies are clearly marked DRAFT or SUPERSEDED for learning purposes.
- Never commit API keys, service-account credentials, or private data.
- Review model names, pricing, quota limits, and access requirements before running the examples.
- Keep the workshop's public participant materials separate from facilitator-only resources.
Created by Decoding Data Science for hands-on AI education and practical enterprise use cases.
Educational use: The exercises and airport scenario are fictional demonstrations, not operational aviation guidance.