<div class="content-intro"><p><img src="https://i.imgur.com/XN94Hg4.png" alt="" width="900" style="max-width: 100%;"></p></div><h3>The Role</h3><p>The Applied Machine Learning Engineer will be a member of <a href="https://himalayas.app/companies/terra-quantum">Terra Quantum</a>'s AI Applied Research team. This team builds and delivers end-to-end machine learning solutions for industrial clients across time series forecasting, optimisation, computer vision, natural language processing, and generative AI. The Engineer will own the classical machine learning craftsmanship that makes those solutions work, from data exploration and feature engineering through to model selection, hyperparameter optimisation, training pipelines, evaluation, and client delivery. A subset of the models built by the team incorporate a quantum layer; the Engineer is expected to treat that layer as one architectural component of an otherwise classical pipeline, and to apply the full toolkit of classical ML methods (including tree-based methods, boosting, deep learning, and classical optimisation) to make hybrid solutions perform reliably on real industrial data.</p><p>The Applied Machine Learning Engineer plays a role in driving excellence within their team. They are not only detail-oriented but also possess a remarkable capacity for enthusiasm. By demonstrating commitment and passion for the mission, they inspire their team members to contribute to making quantum technologies widely accessible and to effect positive change globally.</p><h3>The Responsibilities</h3><p>The Applied Machine Learning Engineer should expect to work in one and supporting in the other areas of the following AI Applied Research Team activities.</p><ul><li>Building and delivering industry machine learning solutions</li><li>Designing end-to-end ML pipelines for client problems in time series, routing and planning, GenAI, natural language processing, computer vision, and predictive modelling</li><li>Choosing the right classical method for the problem (gradient-boosted trees such as XGBoost or LightGBM, random forests, deep neural networks, kernel methods, classical optimisers) based on data characteristics, not framework preference</li><li>Treating the quantum layer (when present) as a constrained component of the model and using classical ML tradecraft (feature engineering, regularisation, training schedules, hyperparameter sweeps) to make the hybrid pipeline work</li><li>Classical machine learning craftsmanship in service of hybrid models</li><li>Doing the unsexy parts of an ML pipeline well: data cleaning, leakage protection, cross-validation design, baseline construction, statistical significance testing</li><li>Designing feature representations that align with the quantum component when one is present, including Fourier-spectrum features and other quantum-aware encodings that classical models can also consume</li><li>Profiling and improving training stability when gradients are noisy or non-standa