{"html":"<p>We&#39;re looking for experienced machine learning researchers with hands-on experience training and improving deep learning models end-to-end, across vision and language. You&#39;ll work on well-scoped empirical open-ended ML research problems.</p>\n<h3>Responsibilities</h3>\n<ul>\n<li>Train image classifiers and generative image models from scratch, and fine-tune open-weight language models.</li>\n<li>Get the most out of limited data, compute, and model-size budgets.</li>\n<li>Make models robust — to adversarial inputs and to adversarial conversations.</li>\n<li>Compress models to meet hard size and latency constraints without sacrificing accuracy.</li>\n<li>Diagnose and resolve training issues.</li>\n</ul>\n<h3>Requirements</h3>\n<p>We are looking for candidates with strong expertise in one or more of the following areas:</p>\n<h4>Adversarial Robustness</h4>\n<p>Experience with:</p>\n<ul>\n<li>Adversarial training of image classifiers (e.g. PGD-based training, TRADES).</li>\n<li>Evaluating robust accuracy under standard threat models (e.g. L∞ attacks, AutoAttack) and avoiding gradient-masking pitfalls.</li>\n<li>Managing the robustness–accuracy trade-off and robust overfitting.</li>\n</ul>\n<h4>Efficient Computer Vision</h4>\n<p>Experience with:</p>\n<ul>\n<li>Training image classifiers end-to-end, especially for fine-grained recognition (many visually similar classes, few examples per class).</li>\n<li>Model compression: quantization, pruning, and knowledge distillation from large teachers into small students.</li>\n<li>Deploying models under hard size or latency budgets (on-device, edge, or embedded settings).</li>\n</ul>\n<h4>Generative Image Modeling</h4>\n<p>Experience with:</p>\n<ul>\n<li>Training image generative models from scratch: diffusion models, GANs, VAEs, or flow-based models.</li>\n<li>Iterating against sample-quality metrics such as FID.</li>\n<li>Training-efficiency tricks that produce good generators quickly and at small parameter counts.</li>\n</ul>\n<h4>LLM Post-Training &amp; Behavioral Robustness</h4>\n<p>Hands-on experience with one or more of:</p>\n<ul>\n<li>Supervised fine-tuning and preference optimisation (DPO, RLHF, RLAIF) of open-weight language models, including building your own datasets via synthetic generation, noisy or weak supervision, and rejection sampling.</li>\n<li>Shaping conversational behaviour over multiple turns: resistance to persuasion and sycophancy, calibrated confidence, and knowing when to accept corrections.</li>\n<li>Alignment-style fine-tuning that changes a specific behaviour while preserving general capability.</li>\n</ul>\n<h4>Multilingual Pre-training</h4>\n<p>Experience with:</p>\n<ul>\n<li>Training multilingual or low-resource-language models from scratch.</li>\n<li>Tokenizer design across scripts and typologically diverse languages.</li>\n<li>Balancing highly unequal per-language data (sampling temperatures, cross-lingual transfer) in data-constrained regimes.</li>\n</ul>\n<h4>Additional Areas of Interest</h4>\n<p>Experience in any of the following is a plus:</p>\n<ul>\n<li>Scaling laws and training-efficiency research.</li>\n<li>Curriculum learning and data ordering.</li>\n<li>Model evaluation: benchmark construction, contamination control, statistically sound comparisons.</li>\n<li>Uncertainty estimation and model calibration.</li>\n<li>Data augmentation and synthetic data for robustness.</li>\n</ul>\n<h3>General Qualifications</h3>\n<ul>\n<li>3+ years of machine learning research experience (PhD research counts toward this requirement).</li>\n<li>Strong experience with PyTorch, JAX, TensorFlow, or similar ML frameworks.</li>\n<li>Degree from a top-100 university, experience at a FAANG or comparable AI company, or an equivalent research track record through publications or impactful open-source contributions.</li>\n</ul>\n<h3>Why Join</h3>\n<ul>\n<li>Work on cutting-edge machine learning research.</li>\n<li>Collaborate with leading AI researchers on challenging, high-impact projects.</li>\n<li>Flexible, project-based work with competitive compensation.</li>\n</ul>"}