Mid/Senior AI LLM Engineer (Full Remote - Netherlands)

<h1>Our Vision &amp; Products</h1><p style="min-height:1.5em"><em><strong>🚀 EverAI — Building the Future of AI Companionship</strong></em></p><p style="min-height:1.5em"><em>One of the Top 15 Largest &amp; Fastest-Growing AI Companies in the World</em></p><p style="min-height:1.5em"><em>50 Million Users in 2 years — Help Us Reach 100M first, 500M next</em></p><p style="min-height:1.5em"></p><p style="min-height:1.5em">At <strong>EverAI</strong>, we’re shaping what it means to connect with AI. With <strong>50 million users and counting</strong>, we're not just building products — we're creating entirely new categories.</p><p style="min-height:1.5em">Our flagship product is the <strong>world’s largest AI companionship platform</strong>, redefining relationships for millions. It is governed by our proprietary moderation system, EverGuard — an internal AI designed to ensure everything we build is <strong>safe, ethical, and human-first</strong>.</p><p style="min-height:1.5em"><em>And we’re only just getting started!</em></p><p style="min-height:1.5em"></p><h1>Our Team</h1><p style="min-height:1.5em">We are an enthusiastic, passionate and hardworking team of ≈ 100 people. Our founding team has strong entrepreneurial experience building and scaling web products from 0 to IPO.</p><p style="min-height:1.5em"></p><p style="min-height:1.5em"><strong>Alexis Soulopoulos </strong>[CEO]</p><p style="min-height:1.5em"><em>• 10+ years in Tech Executive Leadership</em></p><p style="min-height:1.5em"><em>• Co-Founder Mad Paws Holdings (from 0 to IPO)</em></p><p style="min-height:1.5em"><em>• Forbes 30 under 30 + Deloitte TechFast50 ’22 &amp; ‘23</em></p><p style="min-height:1.5em"></p><p style="min-height:1.5em"><strong>Michael Monin</strong> [Co-founder &amp; CTO]</p><p style="min-height:1.5em"><em>• 10+ years as CTO / COO (web2/web3), 3 years in AI/LLM</em></p><p style="min-height:1.5em"><em>• Serial-entrepreneur: MTK Digital (exited / 0-&gt;$20m revenue) and Zipchat (AI Chatbot for E-commerce brands)</em></p><p style="min-height:1.5em"></p><p style="min-height:1.5em"><strong>Thomas Lacroix </strong>[Co-founder &amp; CMO]</p><p style="min-height:1.5em"><em>• 8+ years in Customer Acquisition &amp; E-commerce Growth</em></p><p style="min-height:1.5em"><em>• Serial-entrepreneur: Curatible (sold to Blackstone) and MTK Digital (exited / 0-&gt;$20m revenue)</em></p><p style="min-height:1.5em"></p><p style="min-height:1.5em"><strong>Maruša Fasano </strong>[CFO/Legal]</p><p style="min-height:1.5em"><em>• 25+ years in Finance, Strategy, M&amp;A</em></p><p style="min-height:1.5em"><em>• Ex-CFO/M&amp;A @Curatible (exited to Blackstone)</em></p><p style="min-height:1.5em"><em>• Ex-President of the Board @SotremoSA (exited)</em></p><p style="min-height:1.5em"><em>• Co-founder/CFO @SoftOne (exited)</em></p><p style="min-height:1.5em"></p><h2>Your Role</h2><p style="min-height:1.5em">Most LLM roles stop at experimentation.</p><p style="min-height:1.5em"><strong>Here, the models you improve power experiences for 50M+ users.</strong></p><p style="min-height:1.5em">Most of our traffic runs on <strong>self-hosted, fine-tuned open-weight models</strong>. On our internal roleplay evals, our smaller fine-tuned models can outperform much larger off-the-shelf models. <strong>Your job is to make them better.</strong></p><p style="min-height:1.5em">You’ll join a small AI team led by <strong>Marco, our LLM Tech Lead</strong>, working closely with our CTO, Product, Web and DevOps teams. You’ll work across <strong>data, post-training, evals, and model behavior,</strong> taking ideas from failure analysis to production.</p><p style="min-height:1.5em">The problems are real: <strong>personalization, character consistency, long conversations, better evals, multilingual behavior, and faster training cycles.</strong></p><p style="min-height:1.5em"><strong>If you want to shape model behavior at consumer scale, this is the role.</strong></p><p style="min-height:1.5em"></p><h2>Key Responsibilities</h2><ul style="min-height:1.5em"><li><p style="min-height:1.5em"><strong>Fine-tune open-weight LLMs</strong> using SFT, LoRA/QLoRA and post-training methods such as DPO and GRPO</p></li><li><p style="min-height:1.5em"><strong>Diagnose model failures and turn them into experiments.</strong> Decide whether the answer is better data, post-training, prompting or classifiers</p></li><li><p style="min-height:1.5em"><strong>Build training data and evals</strong> that directly shape what gets trained next</p></li><li><p style="min-height:1.5em"><strong>Improve model behavior</strong> across personalization, consistency, tone, multilingual and group interactions</p></li><li><p style="min-height:1.5em"><strong>Ship to production</strong> and use real user behavior to guide the next iteration</p></li><li><p style="min-height:1.5em"><strong>Speed up the training loop</strong> so experiments are faster, easier to compare, and more reproducible</p></li></ul><p style="min-height:1