Machine Learning / PyTorch Developer

# Machine Learning / PyTorch Developer We are looking for a junior or early-career Machine Learning Developer with a solid understanding of general machine learning concepts and some hands-on experience working with PyTorch. This role is more focused on practical implementation, debugging, and understanding an existing ML training pipeline than on advanced mathematical theory. ## What we're looking for You should be comfortable with the basic structure of a PyTorch training workflow, including: * Working with PyTorch models and understanding how a typical `nn.Module` is structured * Understanding the purpose of the model's `forward()` function and how data flows through a neural network * Working with `Dataset` and `DataLoader`, including understanding how training data is loaded, batched, transformed, and passed into a model * Understanding the basic training loop: model input → forward pass → prediction/output → loss calculation → `loss.backward()` → optimizer step * Understanding PyTorch autograd and the general concept of backpropagation * Familiarity with higher-level training workflows such as `Trainer.train()` or similar training frameworks * Being able to trace the inputs and outputs of different parts of an ML pipeline * Some experience running ML workloads on GPUs * Basic understanding of GPU/device handling such as moving models and tensors between CPU and GPU * Ability to run code in debug mode, set breakpoints, inspect variables/tensors, and follow the execution flow of an unfamiliar codebase * Comfortable reading and modifying Python code You do not need to be an ML research expert. We are primarily looking for someone who understands the fundamentals, has actually run and experimented with PyTorch-based training code, and is comfortable learning by working directly with an existing codebase. ## Typical work The work may involve: * Reading and understanding an existing PyTorch training pipeline * Following how datasets are prepared and converted into model inputs * Debugging model training and data-loading code * Inspecting tensor shapes, model inputs, outputs, and losses * Making small modifications to models, datasets, training configurations, and related Python code * Running experiments on GPU machines * Identifying where problems occur in the training or inference flow ## When applying Please briefly describe: 1. A PyTorch project you have personally worked on 2. What type of model you trained or experimented with 3. Whether you have written or modified a PyTorch `Dataset`, `DataLoader`, or training loop 4. Your experience debugging Python/PyTorch code using breakpoints 5. Your experience running training or inference on GPUs If possible, please include a GitHub repository, code sample, or a short description of something you have implemented yourself.