Case study
dilipsolanki.tech
Product direction & content by Dilip · built with AI-assisted development
Architecture
01
Browser
Minimal JS · reduced-motion aware
02
Edge SSR
TanStack Start renders every route
03
Content layer
Typed TS modules — ventures, work, notes
04
Server function
Zod validation · honeypot
05
Lovable Cloud
Insert-only messages table
What problem needed solving?
My professional identity was spread across LinkedIn and two venture websites. Nothing explained how the engineering work, Dilkresh Studios and Ujjval Panth connect — or gave search engines one clear source of truth for who I am.
What was the context?
A personal platform for recruiters, CTOs, founders and potential collaborators. It had to be credible before anyone scrolls, honest about what's verified, and able to grow with new projects and writing.
What was my contribution?
Positioning, content and product direction, with implementation done through AI-assisted development in Lovable — the workflow itself is one of the things I'm exploring.
Which technologies?
React 19 and TypeScript on TanStack Start with server-side rendering, Tailwind CSS v4 design tokens, Zod for validation and Lovable Cloud for the contact message store.
Which decisions mattered?
- Server rendering for every route so content and metadata are crawlable without JavaScript.
- Typed content modules instead of a database — static information doesn't need one.
- A connected structured-data graph: Person, WebSite, ProfilePage, Article and BreadcrumbList.
- Contact messages validated on client and server, stored in a table the public can write to but never read.
- No portrait. Identity is carried by typography, an original system visualisation and real product screenshots.
What were the trade-offs?
- Publishing a note is a code change, not a CMS click — fine at a deliberate cadence, and easy to revisit.
- SVG and CSS motion instead of WebGL: less spectacle, dramatically lighter on ordinary phones.
- Confidential employment work is shown as practice areas — less concrete, but accurate.
What can be demonstrated?
The live site itself: its routes, sitemap, structured data and working contact form. No traffic or ranking figures are claimed.