What I Built
I’ve always had a love-hate relationship with technology. In college, I majored in computer science and took classes ranging from electrical engineering to human-computer interaction. From soldering transistors on a physical circuit board to designing UI/UX experiences, I’ve touched many layers of the computing stack—and I’m constantly mind-blown by every new piece of the puzzle I learn.
However, my experience as a user of technology before studying it was very different. In middle school and high school, my phone made me feel anxious, stressed, and cynical. I felt lonely on social apps and isolated when I deleted them. I remember summer days in middle school spent alone in my bedroom, watching increasingly extreme dieting videos that YouTube’s recommendations kept feeding me. Back when I had no idea what an algorithm was, I still knew I was being harmed by them.
By my sophomore year of college, I had deleted Instagram and TikTok and turned off YouTube recommendations. While this protected me from harmful content, I also missed important life updates from my close friends and family.
After taking a web development class and learning how to build a basic card layout, I decided to try building my own social app: Lumira.
My goal was simple. I wanted to create a mobile, personal feed of photos just from my friends and family, curated by their genuine interests and sorted by time.
Demo
https://youtube.com/shorts/DwbVU_LFOc0?si=9H5zPovzIbbW8t-i
https://apps.apple.com/us/app/lumira/id6737853449#information
The Comeback Story
Lumira was born out of 3:00 AM manic coding sessions in my college apartment. This was my first fully end-to-end deployed and distributed application—and it was rocky. My code was chaos. My files were unorganized, and I followed no real patterns, but the thing that motivated me to keep going was that it somehow kind of just worked.
I remember the sense of accomplishment I felt the first time I connected to my Firebase backend and saw a photo successfully upload, or the first time I was able to build my code locally and see it displayed on the iOS simulator on my computer.
However, this initial excitement quickly wore off. The inconsistency and lack of structure in my code were making it very difficult to grow my project and making the developer experience very stressful. While I started this project two years ago, I took multiple three-month-long breaks without touching the code, since it seemed that every time I added a new feature, something would break, dependencies would become incompatible, and the amount of time spent versus the amount of progress being made was not worth the effort.
I was fully content with keeping my buggy Lumira on TestFlight forever and moving on with my life.
However, once I started working in industry and began seeing the power of Copilot and agentic coding in general, I was motivated to push Lumira to the finish line.
I had three major breakthroughs that significantly moved this project forward:
- Restructuring and organizing my codebase: I separated concerns across UI components, hooks, stylesheets, and services, which made the app easier to navigate, debug, and scale.
- Defining and prioritizing core features: I clarified my design requirements and focused on what was essential to ship. Earlier versions included features like video uploads and event planning, but I removed them after realizing I couldn’t sustainably maintain them within my time constraints.
- Learning when to leverage AI vs. rely on my own judgment: I developed a better sense of which tasks to delegate to AI, which required deeper review, and when AI wasn’t necessary at all. For my stage of development, manual testing proved most effective—especially after restructuring my codebase. Once I could clearly locate a bug, I was able to fix it efficiently.
My Experience with GitHub Copilot
I view GitHub Copilot as my escalator to the App Store. There were key steps in logic I still had to think through myself to move forward, but Copilot significantly sped up the ride.
I used Copilot in a few core ways:
- Integrating third-party services and libraries: It helped me implement features like Google authentication and quickly adopt new libraries (e.g., drag-and-drop functionality) without getting stuck in documentation.
- Automating repetitive work: I used it to generate scripts for data cleanup and handle repeated patterns like CRUD operations for comments, likes, and photos. Once I built something once, I could provide context and have Copilot scale it across the codebase with consistent structure and formatting.
- Learning and system design support: Copilot acted as a real-time tutor. I asked conceptual questions about caching, client vs. server-side logic, data modeling, and security, which helped me better understand and refine my architecture.
- Debugging partner: My workflow often involved manually testing the app, identifying issues, and then using Copilot (especially in agentic mode) to diagnose and fix bugs quickly.
- UI: I didn’t follow a traditional Figma-driven design process. Once I defined my color palette and core layout (a Polaroid-style card feed and grid view), I relied on Copilot to help implement and iterate on the UI details.
I built Lumira because, after my early experiences with social media applications, I lost much of my trust in them. Through building this app, I gained a deeper understanding of the complexity and incentives driving many existing social networks. I realized this was more of a business and human problem than a technical one, and that neither I nor Copilot could meaningfully change them alone. However, Copilot empowered me to build a simple solution for my specific issue—and, most importantly, to actually ship it.
