AWS Builder Center

Builder Center

Chhaon: Three IIIT-Delhi Students, Four Days, and a Map of Where Delhi Can Find Shade

Delhi’s heat isn’t shared equally. We built a block by block heat map that tells officials which ten places to open and shows residents the nearest cool place on foot, using satellites, open data and AWS.

It's 2 pm and it feels like 43°C

Picture a street in Seelampur, in north-east Delhi, on 25 May 2026. The forecast says it will feel like 43°C. A vendor has nowhere to sit out the afternoon. Five minutes away, a metro station concourse is cool and has drinking water.
He doesn't know it's there. And nobody has told the station it should be open to him.
Every summer, Delhi gets one heat alert for the whole city, but the heat itself is not shared equally. A crowded market with no trees holds heat long after sunset. A park a kilometre away stays cooler. The people hit hardest, like outdoor workers, the elderly and families without cooling at home, often live exactly where the heat stays.
We kept coming back to one question:
If a city could open just ten places as cooling points today, which ten would help the most people?
That question became Chhaon (छाँव), Hindi for shade.

Who we are

We're three students from IIIT-Delhi. We built Chhaon in four days for Environmental Hacks 2026, in the Heat and Water track.
  • Ishaan Gupta, team lead, product design and UI/UX. Split the work, set milestones, designed every screen and co-built the frontend.
  • Harshit Tankha, brand, frontend and user experience. Named the project Chhaon, designed the logo, mapped the user flows and co-built the frontend.
  • Paramveer Oberoi, backend, data and AWS. Built the data pipeline, the heat score and the planning engine, and deployed everything on AWS.
We found the problem together. Then we split into two halves, design on one side and data on the other, and met in the middle.

The idea: four layers, one map

We realised the answer needed four things stacked on top of each other:
  1. Where heat stays. Satellites can see which streets trap heat and which stay cool.
  1. How bad today is. The weather forecast tells you whether today is a normal day or a dangerous one.
  1. Who lives there. A hot empty lot matters less than a hot neighbourhood of thousands.
  1. Where relief is. Metro stations, hospitals, parks, places of worship and water points that could shelter people.
Chhaon combines all four for every 250-metre square of Delhi, 42,387 squares in total. Then it answers two different people.
To a ward officer: "Open these ten places today. Here's how many more at-risk people each one reaches."
To a resident: "It's hot where you are. The nearest cool place is a 6-minute walk, this way."

How we built it

Seeing heat from space

We used two satellites. Landsat measures how hot the ground is. Sentinel-2, which we read straight from the AWS Open Data Registry, shows trees, buildings and water. Because those images are stored in a cloud-friendly format, we could read only the pixels over Delhi instead of downloading whole scenes. In a four-day hackathon, that was the difference between a city-wide map and a map of one neighbourhood.

Making it about today

Satellites tell you where, not when. So every square's score blends 60% of what the satellite sees with 40% of how hot today will feel, from the Open-Meteo forecast. The scale never changes, so the same colour means the same heat on any date, and you can replay a past heatwave to see how bad it really was.

Picking the ten places

Population data from WorldPop tells us how many people live in each square, and we mapped 8,770 possible cooling points from OpenStreetMap. Chhaon then picks places one at a time, always choosing the one that brings the most new at-risk people within a 10-minute walk (800 m).
The result is a ranked list. An officer who can only open four places simply stops at number four.
On 25 May 2026, only 6.4% of high-risk people in Delhi lived within a 10-minute walk of a cooling point. Opening Chhaon's ten picks raises that to 20.5%. That's more than three times as many people, from ten decisions.

The last 500 metres

A list isn't enough at 43°C. A resident needs to know how far, and which way. For the three nearest places, Amazon Location Service gives Chhaon a real walking route and time. If a route is slow, Chhaon falls back to distance and direction rather than leaving someone waiting. In a heatwave, a slow answer is a wrong answer.

Where AWS fits

  • Amazon EC2 runs our API in a Docker container.
  • Amazon ECR stores that container image, so every release ships the same way.
  • Amazon RDS (PostgreSQL) holds all 42,387 squares, the cooling points, daily plans and community reports.
  • Amazon Location Service turns "a place nearby" into "a 6-minute walk".
  • AWS Open Data Registry gives us the Sentinel-2 satellite images.
  • AWS IAM gives our server only the permissions it needs, so no secret keys live in our code.
The choice that made it fast: all the heavy satellite work happens once, ahead of time. When you open the map, Chhaon only reads finished numbers from the database. That's why the heat map for the whole of Delhi loads in about a second.

Designing for a tired person in bright sun

A heat app gets used by people who are hot, tired and often on a cheap phone. That shaped everything we designed.
  • One question, one answer. The public app asks only "Where are you?" and then shows the nearest cool place. You move a satellite map under a fixed pin, or share your location. No typing.
  • In Hindi, not just English. One tap switches every word and every piece of advice.
  • For people without smartphones. Buttons to call 112 and the 1077 Delhi disaster helpline sit right on the main screen.
  • A real control room for officials. The ward screen leads with the gap ("Only 6.4% of high-risk people…"), shows exactly how much each place would help, and lets officers mark which places they've asked to open.

What fought back

  • The prettiest map was the worst map. We redesigned the heat map in soft cream-to-brick tones that looked lovely in our mockups. On real data, all of Delhi turned one orange blur. We went back to green, yellow and red. On a heat map, being readable beats being beautiful.
  • "Failed to fetch." Our first live site loaded, but no data came through: the browser didn't trust the server's security certificate. Lesson: test the real website in a real browser, early.
  • A watermark across our map. A free map style we picked suddenly showed "API KEY REQUIRED" across the screen. We switched to OpenStreetMap with a softened look, with no key needed.
  • The small things. Distances didn't match walking times, so we started measuring along the actual route. Our AWS account defaulted to Sydney, so next time we'd choose Mumbai for a Delhi app.

What we learned

  1. Start with one sharp question. "Which ten places?" kept three people with very different skills pointed the same way.
  1. Do the heavy work once. Precomputing satellite data made the app fast and the cloud bill small.
  1. Design for the hardest moment. Every screen was judged by one test: does this work for someone who's hot, tired and in a hurry?
  1. Open data is a superpower. Satellites, weather, population and maps were all free and open. Our job was to join them.

What's next

We want buildings to be able to accept a request from the ward office, so "asked to open" becomes "open now" on residents' phones. After that: SMS alerts for people in the hottest squares, more Indian languages, and more cities. Every Indian city has satellite coverage.

Tools we used

Next.js, React, TypeScript, Tailwind CSS, Leaflet, tRPC, PostgreSQL and Docker, with Google Earth Engine, Open-Meteo, WorldPop and OpenStreetMap, on Amazon EC2, ECR, RDS, Location Service, IAM and the AWS Open Data Registry. We used Claude (Anthropic) as an AI assistant for design exploration, prototyping and frontend iterations.

Try Chhaon


Back on that street in Seelampur, the vendor has the same phone in his pocket. Now it can tell him the metro station is a 4-minute walk away, and it can tell the city that the station should be open.
That's all chhaon really means: a little shade, in the right place, at the right time.
Ishaan, Harshit and Paramveer Three students from IIIT-Delhi in Okhla, which is, as it happens, one of the 42,387 squares on our map.
Any opinions in this article are those of the individual author and may not reflect the opinions of AWS.
Enjoyed reading this content? Let the author know!

Your likes, comments, shares, and saves help creators reach more builders.

Loading recommendations

Loading article