Weekend Showcase Challenge:PhotoQuest
My fun side project to participate and show case the photo quest having AI Agent running behind the scene under AWS environment.
I Missed the Last Two Challenges. So… Here Goes PhotoQuest.
I had a plan.
I was going to join the challenge.
Then another challenge came.
And somehow, I managed to miss both of the last two challenges. 😅
Not because I didn't have ideas. Quite the opposite—I probably had too many ideas, spent too much time thinking about what would be cool to build, and by the time I was ready to start, the weekend was already over.
So when I saw the latest AWS summer build challenge asking us to take an existing idea, push it further, and make it showcase-ready, I thought:
Okay. Here goes.
No overthinking this time.
I'm building PhotoQuest.
Vision and what it does (the problem it solves and how it works)
I love photography, but there's one problem I think a lot of photographers can relate to:
Sometimes you have a camera.
You have free time.
You even have beautiful weather.
And then you stand there thinking:
"Okay… what should I photograph?"
That's where PhotoQuest comes in.
Instead of asking AI:
"What should I do this weekend?"
PhotoQuest gives you a photography mission.
You choose a mood:
And then you press one button: GIVE ME A QUEST
AI responds
Now you have a reason to grab your camera and actually go somewhere.
How I built it (process, key decisions, challenges)
The fun doesn't stop when you take the photo.
You upload your photograph to PhotoQuest, and an AI photography mentor looks at it in the context of the challenge.
Instead of simply saying:
"This is a good photo."
It gives you a little scorecard:
Creativity: 9/10
Composition: 8/10
Challenge Match: 9/10
Overall: 8.7/10
And then, importantly, it gives you feedback like:
"You found an interesting reflection and turned an ordinary street scene into something surprisingly cinematic."
Followed by one suggestion:
"Next time, try getting lower to exaggerate the reflection."
The goal isn't to replace a photography teacher.
It's to make taking the next photo more tempting.
Building it with Kiro
This is also where I wanted to experiment with Kiro.
Rather than opening an editor and immediately saying:
"Build me an AI photography app."
I started with the idea and used Kiro's Spec-driven development workflow to turn it into something more concrete.
The journey became:
Idea → Requirements → Design → Tasks → Implementation
That was particularly useful because I had to make a decision that I probably wouldn't have made on my own:
Keep it small.
No login.
No social network.
No complicated profiles.
No leaderboard.
No unnecessary database.
Just the core experience:
AWS services used and a brief architecture overview

AWS services used and a brief architecture overview

Amazon Bedrock handles the AI part—both generating photography challenges and evaluating submitted photographs. :let's pick Amazon Luna Lite for the LLM
S3 handles the photographs.
Lambda connects the pieces.
And that's it.
I wanted the architecture to disappear behind the experience rather than become the experience.
Look at these screenshots
- Pick Photo Quest Mood

- The Mood is showed

- Upload photo that was taken

- Bedrock rated my Photographer skill

What I learned across the summer
I learned a lot from other builders previous weekend challenges. Now it is my turn as I decided to build and write this blog. I picked PhotoQuest because I want my hobby turn into reality and want others who share the same interest can enjoy simple app I made and they perhaps have more ideas to have more fun features.
I learned vibe coding with Kiro and explore other areas about Bedrock like picking model to save cost for each type of difference prompts e.g. creating a spec, building frontend, solving bugs.
of course… I couldn't resist security
Being someone who spends a lot of time thinking about security, I also wanted to ask a slightly uncomfortable question:
"What happens when users can upload things and send content to an AI model?"
So before calling it done, I wanted to review things like:
- S3 access
- uploaded file validation
- oversized or malicious files
- CORS
- IAM permissions
- exposed credentials
- API abuse
- prompt injection
- handling unexpected model output
Because building an AI application is fun.
Building one that doesn't accidentally become a security lesson for everyone. I would like to implement this security in the near future. I acknowledge that it is job zero at AWS.
A link to my app or repo
http://photoquest-sitebucket-huv0v0b85cl0.s3-website-us-east-1.amazonaws.com / >> https must be implemented on production. This is just PoC.
A builder who inspired me
I randomly came into @ Marie Otoo Seniagya article and felt like it really matches my hobby, that is why I started building mine.
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article