Machine Learning Engineer

We're seeking a Machine Learning Engineer with a focus on neural and learned data compression. The role involves designing, implementing, and optimising compression models that underpin pdata's core engine. Responsibilities include developing and benchmarking compression pipelines for structured and unstructured data, performing statistical evaluation of rate–distortion and query-accuracy trade-offs, and deploying models into production and pilot environments. You will work directly with the founding team to translate research advances into scalable product capability, maintain code quality and reproducibility, and contribute to technical documentation and white papers. As one of the first technical hires, you will have unusual scope to shape the company's technical direction. Responsibility at Precica grows with contribution and with the company itself — those who build the core of the product will be well placed to lead as we scale. Essential qualifications - Strong foundation in Computer Science: data structures, algorithms, and software engineering principles. - Applied expertise in machine learning, with hands-on experience building and tuning models on real-world data. - In-depth understanding of neural architectures relevant to compression and representation learning. - Proficiency in statistics, probability, and experimental design for rigorous model evaluation. - Fluency in Python and modern ML frameworks (PyTorch, TensorFlow, scikit-learn) and adaptability to work across frameworks. - Experience with large datasets, data pipelines, and version control (Git). - Ability to communicate complex technical concepts clearly and document work thoroughly. - Degree in Computer Science, Mathematics, Engineering, or a related field Desirable qualifications - Direct experience with neural or learned compression: quantisation-aware training, rate–distortion optimisation, or learned transforms. - Postgraduate degree or research experience in machine learning, information theory, signal processing, or data compression. - Background in low-rank or structured representations (e.g. tensor decompositions, sketching, dimensionality reduction). - Experience with time-series modelling and forecasting, particularly in energy or industrial settings. - Familiarity with embedding-based retrieval and video/image representation learning (e.g. CLIP-style models, vector search). - Performance engineering skills: C++, CUDA, or optimisation of inference pipelines for throughput and memory. - Experience deploying models to production (MLOps, containerisation, CI/CD) and with cloud or HPC environments. - Publications, open-source contributions, or patents in a relevant area.