How to present your AI-powered Earth Observation application. Structure your pitch, tell a compelling story, and deliver a demo that captivates your audience.
Your final presentation is not just an academic exercise. It is the moment where your weeks of satellite data wrangling, AI model training, and web development come together into a coherent narrative that others can understand, evaluate, and get excited about.
The difference between a presentation that earns a standing ovation and one that earns polite silence is not technical depth. It is clarity, structure, and emotional resonance.
| Dimension | ❌ Mediocre | ✅ Great |
|---|---|---|
| Opening | "Our project is called TerraVision and we used Sentinel-2 data..." | "Every year, 733 million people face hunger. Our app can see the drought before it happens." |
| Demo | Scrolls through a UI, mumbling "and here you can see..." | Planned 5-step walkthrough with narration for each click |
| Data | "Our model performs well" | "F1 score of 0.87 on 2,400 labeled test tiles, outperforming the baseline by 12%" |
| Slides | Walls of text, 8pt font | One idea per slide, large visuals, minimal text |
| Closing | "So yeah, that's our project. Any questions?" | "With $50K, we could deploy this across East Africa by Q3 2027." |
Tomorrow's room will contain four distinct audiences. Your presentation must speak to all of them simultaneously.
They know your journey. They want to see what you built and how it compares. Show technical ambition and honest reflections on challenges.
They assess methodology, rigor, and learning outcomes. Explain your technical decisions: why this model? Why this data source? What were the trade-offs?
Industry professionals and domain experts. They want real-world applicability. Show market potential, use cases, and scalability.
People who might want to work with you, fund you, or use your app. Give them a reason to approach you after the talk. Make your contact info visible.
Every great tech presentation follows a narrative arc. Your AI-EO application is the hero. The world's problem is the villain. Your audience is the kingdom waiting to be saved.1
The most common mistake in technical presentations is leading with the technology. Your audience does not yet care about your model architecture. They need to feel the urgency of the problem first.
"We built a web application using React, Leaflet, and a ResNet-50 classifier trained on Sentinel-2 imagery to detect deforestation."
The audience is already lost. They don't know why they should care.
"Every few seconds, an area of forest the size of a football field is destroyed. By the time satellite analysts detect it manually, the damage is done. We built a system that detects deforestation in near-real-time."
Urgency first. Solution second. Technology third.
One of the most powerful storytelling frameworks is the contrast between the world before your solution and the world after. Nancy Duarte calls this the "sparkline," alternating between "what is" and "what could be" to create emotional momentum.1
| Dimension | 🔴 Before Our App | 🟢 After Our App |
|---|---|---|
| Detection speed | Manual analysis: 2-3 weeks | Automated alerts: under 48 hours |
| Coverage | Sample-based monitoring (5% of area) | Wall-to-wall coverage via satellite |
| Cost | $15,000 per field survey campaign | $200/month for cloud processing |
| Accessibility | Requires GIS expertise | Any user with a web browser |
The before/after framework works because it creates a contrast gap. The audience's brain naturally wants to resolve the tension, which makes your solution feel inevitable rather than arbitrary.
Precision builds credibility. Vagueness destroys it. Your evaluators and potential collaborators want to know exactly what your system achieves.
| ❌ Vague | ✅ Concrete |
|---|---|
| "Helps farmers" | "25% reduction in crop loss through early drought alerts" |
| "Fast detection" | "Processes 500 km² in 3.2 seconds on a single GPU" |
| "High accuracy" | "F1 = 0.87, precision = 0.91, recall = 0.83 on 2,400 test tiles" |
| "Easy to use" | "3 clicks from login to risk map. Average task time: 12 seconds." |
| "Scalable solution" | "Currently monitoring 8 regions. Architecture supports 200+ with no code changes." |
In Earth Observation, you have a superpower most tech presenters lack: your data is inherently visual. Satellite images, false-color composites, classification maps, and change detection overlays are more compelling than any bullet point.
What is the FIRST thing your presentation should establish?
You have exactly 10 minutes. Every second counts. Here is a proven structure that balances narrative, demonstration, and technical depth.3
Hook the audience. State the problem, its scale, and human cost. Use a startling statistic or a personal story. End with: "What if we could change this?"
Introduce your application. Describe the core concept in one sentence. Show a hero screenshot. Explain the data sources (Sentinel-2, Landsat, etc.) and AI approach at a high level.
This is your centerpiece. Walk through 3-5 key interactions in your app. Narrate every click. Show real data, real results. This is where you prove it works.
One slide with your system diagram. Frontend, backend, data pipeline, AI model. Mention key technologies. Keep it to 60 seconds.
Quantify impact. Discuss limitations honestly. Present your roadmap: next 3 months, next year. End with a call to action.
Wrap up with your "CEO number" and contact info. Thank the audience. Transition smoothly into Q&A. Aim to finish at 9:30, giving you a 30-second safety buffer.
Your opening must answer one question: "Why should I care?" The evaluator's attention is at its peak in the first 60 seconds. Do not waste it on introductions or agenda slides.
Now introduce your application with a one-sentence value proposition:
After the value proposition, show your hero screenshot: the single most impressive view of your running application. Then briefly explain your data pipeline and AI approach at a conceptual level (save the technical details for the architecture slide).
The live demo is the moment of truth. This is where you prove that your application is not just a concept but a working product. It is also the riskiest part of your presentation.
Plan every single click in advance. Write a script that looks like this:
One slide. One diagram. Show the data flow from satellite to user. Your architecture slide should answer three questions:
Close strong. Quantify your impact, acknowledge limitations honestly, and show ambition:
Your slides are not your script. They are your visual backdrop. The audience should be listening to you, not reading your slides. Follow these rules from presentation design research.4 Mayer's Cognitive Theory of Multimedia Learning confirms that presenting words and visuals together (without redundant text) leads to deeper understanding.7
Each slide should communicate exactly one concept. If you need a second idea, make a second slide. Slides are free.
Your slide title should be the takeaway, not a label. "Drought Detected 3 Weeks Early" beats "Model Results Summary."
If you cannot read the text from 3 meters away, it is too small. Minimum font size: 24pt for body text, 36pt for titles.
Use your app's color scheme in your slides. If your app uses emerald green and dark backgrounds, your deck should match. This reinforces brand identity.
Charts in presentations are not academic figures. They are arguments. Every chart should make exactly one point, and that point should be the chart's title.5 Cole Nussbaumer Knaflic's framework emphasizes identifying the "so what" in your data before choosing a chart type.8
| Principle | Example |
|---|---|
| Title = Conclusion | Instead of "Accuracy by Model," write "ResNet-50 Outperforms All Baselines" |
| Highlight the key bar | Color your model's bar in your accent color; make all others gray |
| Annotate the "so what" | Add a callout: "+12% over baseline" with an arrow pointing to the gap |
| Remove chartjunk | No gridlines, no borders, no 3D effects, no legends if only 2 series |
| Use consistent colors | Green = your model. Gray = baselines. Red = failure cases. Every chart. |
Beyond your presentation slides, your project lives on through the AS26 Project Showcase. This is your portfolio entry, visible to future employers, collaborators, and ISU alumni worldwide.
What is the maximum number of key points you should have on a single presentation slide?
Rehearsal is not optional. Research on presentation performance shows that speakers who rehearse with structured feedback score significantly higher on clarity, confidence, and audience engagement.6 A systematic review of 52 studies identified "opportunity to practice" and "intensity and timing of feedback" as two of seven key design principles for developing oral presentation competence.10
Before you leave today, make sure every item on this list is either done or has a clear plan. Check off each item as you complete it:
Data Visualization Pioneer
Hans Rosling co-founded Gapminder and revolutionized how the world understands global development data. His animated bubble charts and TED talks showed that compelling data visualization can change minds and fight ignorance. His book 'Factfulness' remains essential reading for anyone presenting data.
Earth Observation provides a macroscopic view of environmental trends, but its true power lies in downscaling this data to affect local policy and design, such as urban planning and sustainable workplaces.
Your startup needs to establish a new hybrid work hub. You must balance employee commute times, environmental impact (using the IPAT equation), and existing green infrastructure.
What was your biggest takeaway from this session, and how does it apply to the TERRA project? Write your response below. Your instructor will review this to track your progress.