ML/AI engineer portfolio website, built by Claude Haiku 5.5

<!doctype html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>Mira Kovač — Senior Machine Learning Engineer</title>
<meta name="description" content="CV of Mira Kovač, Senior Machine Learning Engineer based in Berlin.">
<style>
  :root {
    --bg: #0b0f0c;
    --fg: #cdd7cf;
    --muted: #7b8a80;
    --rule: #1b251e;
    --edge: #2e3d33;
    --green: #3ddc84;
    --font: ui-monospace, "SF Mono", Menlo, Consolas, "Liberation Mono", monospace;
    --s1: 8px;
    --s2: 16px;
    --s3: 24px;
    --s4: 32px;
    --s5: 48px;
    --s6: 64px;
  }

  * { box-sizing: border-box; }

  html { background: var(--bg); }

  body {
    margin: 0;
    background: var(--bg);
    color: var(--fg);
    font: 15px/1.65 var(--font);
    -webkit-text-size-adjust: 100%;
  }

  ::selection { background: var(--green); color: var(--bg); }

  a { color: inherit; }
  a:focus-visible { outline: 1px solid var(--green); outline-offset: 3px; }

  .wrap {
    max-width: 720px;
    margin: 0 auto;
    padding: 0 var(--s2);
  }

  /* Header */
  header { padding: var(--s6) 0 var(--s5); }

  .prompt {
    margin: 0;
    color: var(--muted);
    overflow-wrap: anywhere;
  }
  .prompt b { color: var(--green); font-weight: 400; }
  .prompt.end { margin-top: var(--s5); }

  .caret {
    display: inline-block;
    width: 0.6em;
    height: 1em;
    margin-left: 2px;
    vertical-align: -0.15em;
    background: var(--green);
    animation: blink 1.1s steps(1) infinite;
  }
  @keyframes blink { 50% { opacity: 0; } }
  @media (prefers-reduced-motion: reduce) { .caret { animation: none; } }

  h1 {
    margin: var(--s4) 0 0;
    font-size: clamp(1.75rem, 6vw, 2.5rem);
    line-height: 1.15;
    font-weight: 700;
    letter-spacing: -0.02em;
  }

  .role {
    margin: var(--s1) 0 0;
    color: var(--green);
  }

  .summary {
    margin: var(--s3) 0 0;
    max-width: 58ch;
  }

  .menu {
    display: flex;
    flex-wrap: wrap;
    gap: var(--s1) var(--s3);
    margin: var(--s4) 0 0;
    padding: 0;
    list-style: none;
    font-size: 0.875rem;
  }
  .menu a { color: var(--muted); text-decoration: none; }
  .menu a:hover { color: var(--green); }

  /* Sections */
  section {
    padding: var(--s6) 0;
    border-top: 1px solid var(--rule);
  }

  h2 {
    margin: 0 0 var(--s4);
    font-size: 0.875rem;
    font-weight: 400;
    color: var(--muted);
  }
  h2 .sym { color: var(--green); margin-right: var(--s1); }

  /* Experience timeline */
  .timeline {
    list-style: none;
    margin: 0;
    padding: 0;
  }
  .timeline > li {
    display: grid;
    grid-template-columns: 112px 1fr;
    gap: 0 var(--s3);
  }
  .timeline > li + li { margin-top: var(--s5); }

  .year {
    color: var(--green);
    white-space: nowrap;
  }

  .entry {
    position: relative;
    padding-left: var(--s3);
    border-left: 1px solid var(--edge);
  }
  .entry::before {
    content: "";
    position: absolute;
    left: -4px;
    top: 9px;
    width: 7px;
    height: 7px;
    background: var(--green);
  }

  .entry h3 {
    margin: 0;
    font-size: 1rem;
    font-weight: 700;
    line-height: 1.5;
  }
  .org {
    display: block;
    margin-top: 2px;
    font-weight: 400;
    color: var(--muted);
  }

  .bullets {
    list-style: none;
    margin: var(--s2) 0 0;
    padding: 0;
  }
  .bullets li {
    position: relative;
    padding-left: var(--s3);
  }
  .bullets li::before {
    content: ">";
    position: absolute;
    left: 0;
    color: var(--muted);
  }
  .bullets li + li { margin-top: var(--s2); }

  /* Skills */
  .chips {
    list-style: none;
    margin: 0;
    padding: 0;
    display: flex;
    flex-wrap: wrap;
    gap: var(--s1);
  }
  .chips li {
    padding: 4px 12px;
    border: 1px solid var(--edge);
    line-height: 1.5;
  }

  /* Projects and publication */
  .rows {
    list-style: none;
    margin: 0;
    padding: 0;
  }
  .rows li {
    display: grid;
    grid-template-columns: 180px 1fr;
    gap: 0 var(--s3);
    padding: var(--s2) 0;
    border-top: 1px solid var(--rule);
  }
  .rows li:last-child { border-bottom: 1px solid var(--rule); }
  .rows .name { color: var(--green); }

  .pub {
    margin: 0;
    padding: var(--s2) 0;
    border-top: 1px solid var(--rule);
    border-bottom: 1px solid var(--rule);
  }
  .pub em { font-style: normal; color: var(--green); }

  /* Contact row */
  .contact {
    display: flex;
    flex-wrap: wrap;
    gap: var(--s1) var(--s4);
    margin: 0;
    padding: 0;
    list-style: none;
  }
  .contact a { text-decoration: none; border-bottom: 1px solid var(--edge); }
  .contact a:hover { color: var(--green); border-color: var(--green); }
  .contact .label { color: var(--muted); }

  footer {
    padding: var(--s5) 0;
    border-top: 1px solid var(--rule);
    font-size: 0.8125rem;
    color: var(--muted);
  }

  /* Phone width */
  @media (max-width: 560px) {
    header { padding: var(--s5) 0 var(--s4); }

    .timeline > li { grid-template-columns: 1fr; gap: var(--s1) 0; }
    .year { padding-left: var(--s3); }

    .rows li { grid-template-columns: 1fr; gap: var(--s1) 0; }
  }
</style>
</head>
<body>
<div class="wrap">

  <header>
    <p class="prompt"><b>mira@berlin</b>:~$ cat cv.md</p>
    <h1>Mira Kovač</h1>
    <p class="role">Senior Machine Learning Engineer · Berlin</p>
    <p class="summary">Nine years building models that have to survive contact with operations: demand forecasting, document NLP for logistics, and the MLOps that keeps them honest after launch.</p>

    <ul class="menu">
      <li><a href="#experience">experience</a></li>
      <li><a href="#skills">skills</a></li>
      <li><a href="#projects">projects</a></li>
      <li><a href="#publications">publications</a></li>
      <li><a href="#contact">contact</a></li>
    </ul>

    <p class="prompt end"><b>mira@berlin</b>:~$ <span class="caret" aria-hidden="true"></span></p>
  </header>

  <section id="experience">
    <h2><span class="sym">$</span>ls experience/</h2>

    <ol class="timeline">
      <li>
        <div class="year">2022 — now</div>
        <div class="entry">
          <h3>Senior Machine Learning Engineer<span class="org">Ostkante Mobility, Berlin</span></h3>
          <ul class="bullets">
            <li>Own the demand-forecasting stack behind fleet rebalancing across 14 cities. Replaced 40 per-station LightGBM models with one hierarchical model using learned station embeddings; weekly MAPE fell from 18.4% to 11.2% on held-out stations.</li>
            <li>Found a label-leakage bug in the backtest, where future bookings were leaking into lag features. Fixing it lowered the reported accuracy by about three points, and every model now reads features from a point-in-time-correct store.</li>
            <li>Lead a team of three. Built drift monitoring on input PSI and output calibration; alerts fire when a city's error runs above its 14-day baseline, which caught two upstream sensor outages this year.</li>
          </ul>
        </div>
      </li>

      <li>
        <div class="year">2019 — 2022</div>
        <div class="entry">
          <h3>Machine Learning Engineer<span class="org">Kiefernhof Logistik, Berlin</span></h3>
          <ul class="bullets">
            <li>Shipped a transformer-based address-normalisation model for the parcel-sorting pipeline. Manual re-routing dropped from 6.1% to 1.7% of parcels after rollout to four hubs.</li>
            <li>Quantised the model to INT8 with ONNX Runtime. p95 CPU latency went from 47 ms to 9 ms, so the hub edge boxes run it without a GPU.</li>
            <li>Set up active learning on low-confidence splits with the operations team. The labelled set doubled in one quarter without adding annotation headcount.</li>
          </ul>
        </div>
      </li>

      <li>
        <div class="year">2017 — 2019</div>
        <div class="entry">
          <h3>Data Scientist<span class="org">Halden Energy Analytics, Berlin</span></h3>
          <ul class="bullets">
            <li>Built load-forecasting models for two regional grid operators, using gradient-boosted trees over STL-decomposed demand. Beat the incumbent ARIMA baseline at 9 of 12 substations in the year-end review.</li>
            <li>Wrote the smart-meter ingestion checks (gap detection, DST and timezone handling) that became the team's reference pipeline.</li>
            <li>Automated weekly retraining on Airflow with experiments tracked in MLflow, turning a two-day manual refresh into an overnight job.</li>
          </ul>
        </div>
      </li>
    </ol>
  </section>

  <section id="skills">
    <h2><span class="sym">$</span>cat skills</h2>
    <ul class="chips">
      <li>Python</li>
      <li>PyTorch</li>
      <li>LightGBM</li>
      <li>Forecasting</li>
      <li>MLflow</li>
      <li>Kubeflow</li>
      <li>ONNX Runtime</li>
      <li>SQL</li>
    </ul>
  </section>

  <section id="projects">
    <h2><span class="sym">$</span>ls projects/</h2>
    <ul class="rows">
      <li>
        <span class="name">./wordle-solver</span>
        <span>Entropy-maximising Wordle guesser that averages 3.6 guesses across the full answer list.</span>
      </li>
      <li>
        <span class="name">./onnx-bench</span>
        <span>Latency harness comparing ONNX Runtime, OpenVINO and TFLite on small vision models across CPU classes.</span>
      </li>
      <li>
        <span class="name">./ubahn-tiles</span>
        <span>Vector-tile map of Berlin's U-Bahn network, rendered straight from GTFS in under 400 lines of Python.</span>
      </li>
    </ul>
  </section>

  <section id="publications">
    <h2><span class="sym">$</span>cat publications</h2>
    <p class="pub">
      Kovač, M. (2021). &ldquo;Hierarchical station embeddings for sparse demand forecasting.&rdquo;
      <em>Proceedings of the Nordic Workshop on Applied Time Series.</em>
    </p>
  </section>

  <section id="contact">
    <h2><span class="sym">$</span>echo contact</h2>
    <ul class="contact">
      <li><span class="label">email </span><a href="mailto:mira@example.com">mira@example.com</a></li>
      <li><span class="label">github </span>mira-kovac</li>
      <li><span class="label">based in </span>Berlin, DE</li>
    </ul>
  </section>

  <footer>
    Fictional sample CV. The persona, employers and projects are invented.
  </footer>

</div>
</body>
</html>