About The Role
The role focuses on taking machine learning models from research notebooks to reliable, monitored production systems. This means owning training pipelines, deployment infrastructure, and inference services where latency, cost, and accuracy are all engineering constraints — not afterthoughts.
This is a hands-on position on a small, senior team. The work spans classical ML and deep learning, with increasing exposure to LLM-based systems, and directly shapes how models perform for real users at scale.
Key Responsibilities
Design, train, and evaluate ML models for production use cases spanning tabular, NLP, and increasingly LLM-driven workloads
Build and maintain training and feature pipelines in Python and PySpark, ensuring parity between offline experimentation and online serving
Deploy and operate models using Docker, Kubernetes, and cloud ML platforms (AWS SageMaker, GCP Vertex AI, or Azure ML), including canary rollouts and rollback procedures
Instrument model monitoring for data drift, latency, and performance regression using tools like Evidently, Prometheus, or Grafana
Optimize inference performance: batching, quantization, ONNX conversion, and GPU utilization tuning
Collaborate with data engineers and product teams to translate business requirements into well-scoped ML problems with measurable success criteria
Contribute to team engineering standards through code reviews, design docs, and reproducible experimentation practices
What We Are Looking For
3–6 years of experience in machine learning engineering, MLOps, or applied data science with production deployments
Strong Python engineering skills; comfort with software engineering practices like testing, CI/CD, and code review
Hands-on proficiency with PyTorch or TensorFlow, plus scikit-learn for classical ML baselines
Experience deploying and operating models on cloud infrastructure (AWS, GCP, or Azure) with Kubernetes or equivalent orchestration
Solid ML fundamentals: evaluation methodology, regularization, handling class imbalance, and overfitting in real datasets
Master's degree in CS, Statistics, Engineering, or equivalent practical experience
Bonus: Experience fine-tuning or serving LLMs (LoRA, vLLM, Triton), building RAG systems, or working with feature stores like Feast or Tecton