Machine Learning Engineer

Job Title: Data Science Mentor Expertise– Causal Machine Learning (PyTorch / PyTorch Lightning) Job Type: Contract, Part-Time, Remote Duration: [2–3] months (extendable) Commitment:3 hr per day,15 -20 hours weekly Job Summary We are looking for an experienced Machine Learning professional to provide ongoing mentorship and technical guidance for a causal discovery project. The project uses a Transformer-based Variational Autoencoder combined with a learnable causal graph, trained in two stages using PyTorch Lightning. The ideal candidate has strong deep learning and causal inference expertise and can explain complex concepts clearly and practically. Key Responsibilities - Provide regular one-on-one mentoring sessions to explain model architecture, training workflows, and underlying concepts - Guide understanding of Transformer VAE components, including encoder, decoder, positional embeddings, latent representations, and KL divergence - Explain causal graph learning techniques, including learnable adjacency matrices, acyclicity constraints (NOTEARS-style), augmented Lagrangian optimization, and sparsity and entropy penalties - Support understanding of two-stage training in PyTorch Lightning, including callbacks, mixed precision training, and gradient clipping - Assist with debugging training issues such as NaN losses, unstable convergence, and numerical instability - Guide graph analysis using NetworkX, including DAG validation, cycle detection, and path enumeration - Explain causal effect estimation methods, including ATE, Bayesian ATE, adjusted causal effects, and counterfactual analysis - Advise on time-series preprocessing and stationarity testing ADF using pandas and statsmodels - Provide guidance on experiment tracking and comparison using MLflow - Review approaches to tasks and recommend best practices, papers, and learning resources - Be available for quick questions via chat between sessions Required Skills and Qualifications - 5+ years of hands-on experience with Python and PyTorch - Strong working knowledge of PyTorch Lightning - Practical experience with Variational Autoencoders and Transformer architectures - Solid understanding of causal discovery and causal inference NOTEARS, DAG-GNN, or similar methods - Knowledge of constrained optimization, including Lagrangian methods - Experience with NumPy, pandas, and statsmodels - Familiarity with NetworkX and graph algorithms - Experience training models on GPU CUDA including mixed precision - Excellent communication skills and the ability to explain technical concepts simply Preferred Qualifications - Experience with MLflow or similar experiment tracking tools - Background in time-series causal modeling - Knowledge of Bayesian methods for treatment effect estimation - Master's or PhD in Machine Learning, Statistics, Computer Science, or a related field - Prior experience mentoring, teaching, or providing technical job support