Personal Project
Meetups
Meetups is a rendering-strategy study using Next.js SSG with ISR: MongoDB-backed data, build-time pages for SEO, regenerated when listings change.
Next.jsMongoDBTailwind CSSSSG
View Details—·Personal Project
CaptionForge
I built CaptionForge as my first shipped AI product. Upload an image, get platform-ready captions and SEO hashtags from OpenAI in one click. Next.js 15, React 19, Tailwind 4.
Nov 2025
Overview
CaptionForge started as a small side project scoped around one moment: upload to publish-ready output. OpenAI reads visual context and returns caption variants plus hashtag sets for Instagram, LinkedIn, or Twitter. Skeleton loading and OG-optimized landing pages keep first-use latency trustworthy. This is where I learned prompt engineering and API cost awareness before building agent systems in MedLens+.
Role
Solo product engineer from idea through landing page, OpenAI integration, and Vercel deployment.
Scope
Image upload, OpenAI vision prompt for caption and hashtag generation, copy-to-clipboard UX, and SEO-optimized marketing pages with dynamic OG images.
Technologies
Next.jsReact 19OpenAI APITailwind CSSTypeScript
Key Learnings
- How to scope an AI product around one high-value moment, upload to publish-ready output
- Prompt engineering for platform-specific caption tone without over-engineering the UI
- Marketing-site SEO patterns (OG images, sitemaps) that help micro-SaaS get discovered organically
Challenges & Solutions
- Managing OpenAI latency so users trust the product on first try
- Writing prompts that feel native to each social platform's voice
- Balancing API cost awareness with a frictionless free-tier experience
Business Impact
- Caption writing drops from 15+ minutes to under 30 seconds per post for daily creators.
- Single-screen demo flow (upload, generate, copy) is easy to show in client conversations.
- Early proof of shipping AI features users can touch, not just architecture diagrams.
Outcome
- Live product at caption-forge-ai.vercel.app with polished generate flow.
- Foundation patterns for OpenAI integration reused in MedLens+ and hackathon agent work.
- Learned to balance API cost awareness with a frictionless first-use experience.
Related Projects
Personal Project
Metaversus UI/UX
I built Metaversus as a motion and performance sandbox for a fictional metaverse agency, with scroll-triggered Framer Motion sequences, page transitions, and layered effects in Next.js at 60fps.
Next.jsFramer MotionTailwind CSSCSS Modules
View Details
Personal Project
MedLens+
My ongoing side build for lab report chaos. MedLens+ extracts CBC, LFT, and thyroid values from clinic PDFs into longitudinal timelines. Agents cite ranges from the uploaded report. They never diagnose.
Next.jsFastAPIPythonPostgreSQL
View DetailsWant similar delivery?Schedule a Call