diff --git a/scrapper/README.md b/scrapper/README.md
index 7d901b4..92203c1 100644
--- a/scrapper/README.md
+++ b/scrapper/README.md
@@ -4,6 +4,13 @@ Scrapes job sites once a day, stores results in SQLite (`data/jobs.sqlite`), and
 
 No runtime dependencies. It uses Bun's built-in SQLite, HTTP server and HTMLRewriter.
 
+## Sources
+
+| Source | How | Notes |
+|---|---|---|
+| slo-tech.com (Ljubljana) | HTML scrape | ~20 newest jobs |
+| startup.si | HTML scrape (no public API or feed) | Site lists every posting since 2019 on one page, so only the last 90 days are kept (`startupSi(days)` in `src/scrapers/index.ts`) |
+
 ## Run
 
 ```sh
diff --git a/scrapper/src/scrapers/http.ts b/scrapper/src/scrapers/http.ts
index 97e83d5..65a771c 100644
--- a/scrapper/src/scrapers/http.ts
+++ b/scrapper/src/scrapers/http.ts
@@ -29,6 +29,8 @@ export async function fetchHtml(url: string): Promise<string> {
 
 const NAMED_ENTITIES: Record<string, string> = {
   amp: "&", lt: "<", gt: ">", quot: '"', apos: "'", nbsp: " ", raquo: "»", laquo: "«",
+  scaron: "š", Scaron: "Š", zcaron: "ž", Zcaron: "Ž", ccaron: "č", Ccaron: "Č",
+  ndash: "–", mdash: "—", lsquo: "‘", rsquo: "’", ldquo: "“", rdquo: "”", bdquo: "„", hellip: "…",
 };
 
 /** Decode HTML entities (HTMLRewriter text chunks are not decoded). */
@@ -38,7 +40,7 @@ export function decodeEntities(s: string): string {
       const code = e[1] === "x" || e[1] === "X" ? parseInt(e.slice(2), 16) : parseInt(e.slice(1), 10);
       return Number.isFinite(code) ? String.fromCodePoint(code) : m;
     }
-    return NAMED_ENTITIES[e.toLowerCase()] ?? m;
+    return NAMED_ENTITIES[e] ?? NAMED_ENTITIES[e.toLowerCase()] ?? m;
   });
 }
 
diff --git a/scrapper/src/scrapers/index.ts b/scrapper/src/scrapers/index.ts
index e5afb1a..ba1a3a0 100644
--- a/scrapper/src/scrapers/index.ts
+++ b/scrapper/src/scrapers/index.ts
@@ -1,4 +1,5 @@
 import { sloTech } from "./slotech";
+import { startupSi } from "./startupsi";
 import type { Scraper } from "./types";
 
 /**
@@ -8,6 +9,7 @@ import type { Scraper } from "./types";
  */
 export const scrapers: Scraper[] = [
   sloTech("Ljubljana"),
+  startupSi(),
 ];
 
 export type { Scraper, ScrapedJob } from "./types";
diff --git a/scrapper/src/scrapers/startupsi.ts b/scrapper/src/scrapers/startupsi.ts
new file mode 100644
index 0000000..b183b9b
--- /dev/null
+++ b/scrapper/src/scrapers/startupsi.ts
@@ -0,0 +1,94 @@
+import { clean, fetchHtml } from "./http";
+import type { ScrapedJob, Scraper } from "./types";
+
+// startup.si has no public jobs API or feed, so we parse the listing page. It holds every posting
+// on a single page (no pagination), going back years, so old postings are filtered out by age.
+const LIST_URL = "https://www.startup.si/sl-si/za-startupe/oglasi/zaposlitve";
+
+const MONTHS = ["januar", "februar", "marec", "april", "maj", "junij", "julij", "avgust", "september", "oktober", "november", "december"];
+
+/** "21. september 2026" -> ISO date */
+function parseSlDate(s: string): string | undefined {
+  const m = /(\d{1,2})\.\s*([a-zčšž]+)\s+(\d{4})/i.exec(s);
+  const month = m ? MONTHS.indexOf(m[2]!.toLowerCase()) : -1;
+  if (!m || month < 0) return undefined;
+  return new Date(Date.UTC(Number(m[3]), month, Number(m[1]))).toISOString();
+}
+
+/** Links on this site are hand-entered: some are empty, relative, or missing the scheme. */
+function normalizeUrl(href: string): string {
+  const h = clean(href);
+  if (!h) return LIST_URL;
+  if (/^www\./i.test(h)) return `https://${h}`;
+  try {
+    return new URL(h, LIST_URL + "/").href;
+  } catch {
+    return LIST_URL;
+  }
+}
+
+/** Scraper for startup.si job ads. Only postings from the last `maxAgeDays` days are returned. */
+export function startupSi(maxAgeDays = 90): Scraper {
+  return {
+    id: "startup-si",
+    name: "Startup.si",
+    async scrape() {
+      const jobs = await parseStartupSi(await fetchHtml(LIST_URL));
+      if (jobs.length === 0) throw new Error("no job cards found (site layout changed?)");
+      const cutoff = Date.now() - maxAgeDays * 86400_000;
+      return jobs.filter((j) => !j.postedAt || new Date(j.postedAt).getTime() >= cutoff);
+    },
+  };
+}
+
+export async function parseStartupSi(html: string): Promise<ScrapedJob[]> {
+  type Draft = { id: string; title: string; date: string; href?: string };
+  const drafts: Draft[] = [];
+  let cur: Draft | undefined;
+  let lastHref = "";
+
+  await new HTMLRewriter()
+    .on(".advertisment-card", {
+      element() {
+        cur = { id: "", title: "", date: "" };
+        drafts.push(cur);
+      },
+    })
+    .on(".advertisment-card .right > .title", {
+      text(t) {
+        if (cur) cur.title += t.text;
+      },
+    })
+    .on(".advertisment-card .title button", {
+      element(el) {
+        if (cur) cur.id = (el.getAttribute("data-target") ?? "").replace(/^#description/, "");
+      },
+    })
+    .on(".advertisment-card .published-info b", {
+      text(t) {
+        if (cur) cur.date += t.text;
+      },
+    })
+    // The "preberi več" (read more) link is an <a> wrapping span.read-more; descriptions can contain other links.
+    .on(".advertisment-card a", {
+      element(el) {
+        lastHref = el.getAttribute("href") ?? "";
+      },
+    })
+    .on(".advertisment-card span.read-more", {
+      element() {
+        if (cur) cur.href = lastHref;
+      },
+    })
+    .transform(new Response(html))
+    .text();
+
+  return drafts
+    .filter((d) => d.id && clean(d.title))
+    .map((d) => ({
+      externalId: d.id,
+      title: clean(d.title),
+      url: normalizeUrl(d.href ?? ""),
+      postedAt: parseSlDate(clean(d.date)),
+    }));
+}