My portfolio was frozen for five years.
One evening, it was mine again

The first hours and days

1. Mar 29 13:48

   The old site has no export button. So the first instruction I gave the agent was:

   "Download my entire Adobe Portfolio site, all pages, all images, all assets, into a clean organized folder on my computer."

   

2. Mar 29 354

   lines in a script that went through 19 project pages and four general ones. The old site hides every image until you scroll to it, so we looked for each one in three different places in the page code.

   for img in soup.find_all("img"):
       for attr in ("src", "data-src", "data-image"):
           src = img.get(attr, "").strip()
           if src and not src.startswith("data:"):
               image_urls.add(urljoin(base, src))

3. Mar 29 3

   platforms the videos lived on: Facebook, YouTube and Vimeo. Each video came down with a separate tool, and whatever failed was saved as a link.

   result = subprocess.run(
       [
           "python3", "-m", "yt_dlp",
           "--no-playlist",
           "-o", str(dest_dir / "%(id)s.%(ext)s"),
           url,
       ],

4. Mar 29 19

   videos and prototypes that cannot be downloaded at all: 17 in Adobe’s player on the post production page, and two in Adobe XD. They were saved as links and came into the new site as embedded windows.

   elif "adobe.io" in src or "xd.adobe.com" in src:
       reference_urls.add(src)

5. Mar 29 13:58

   Ten minutes in: 19 project folders and 99 files. iCount came back empty, because the old site had locked it with a password, so we built it separately later.

   {
     "slug": "leadershipaticount",
     "title": "Password Protected Page",
     "url": "https://www.meravish.com/leadershipaticount",
     "media_count": 0
   }

   

6. Mar 29 26MB

   for a single GIF in the motion project. That project alone held 22 GIFs, 53MB that every visitor downloads.

   26M   539a8e1f-c10c-41ff-836f-3879b7565906_rw_1200.gif
   3.5M  a4ca336f-fa48-410e-bbf7-98848459e70c_rw_600.gif
   3.2M  2c11e7f0-510c-4d6a-9119-740ee2947faf_rw_1200.gif

7. Apr 21 06:55

   1,044 lines in the script that built the site from the folder: a home page, 18 project pages and a work with me page. Every project asked for something else: iCount was built separately, and a Hebrew tagline was swapped for the description from the list.

   def build_project(slug, year, title, desc_short):
       if slug == "leadershipaticount":
           build_icount_page()
           return
       ...
       # If tagline contains Hebrew characters, fall back to the PROJECTS-list desc
       if any('\u0590' <= c <= '\u05ff' for c in (tagline or "")):
           tagline = desc_short

8. Apr 21 x32

   is the size the old site signed its thumbnails at, and they came out blurry. We tried to upscale them, then gave up and took each project’s original image from the folder.

   def hires_thumb_url(url):
       """Upgrade MyPortfolio CDN thumbnail from x32 → x800."""
       import re
       return re.sub(r'x32\.(png|gif|jpg)', r'x800.\1', url)
       ...
       return False  # Always use first_local_media fallback instead

9. 2 days

   after the site went up, Netlify suspended it for too much traffic. Vercel got stuck at sign in with version 52 of its tool. Only on the third try, on Cloudflare, did it stay up, and the address moved over.

   registrar     GoDaddynameservers   dora.ns.cloudflare.com              odin.ns.cloudflare.comA record      Netlify   (deleted)​$ npx wrangler@latest deploy --name meravish

10. 2 hours

    and meravish.com still showed an error page. Netlify was holding my address where I couldn’t see it. Removing the domain wasn’t enough; only deleting my account there released it.

In short, here is how it felt: