ML Engineer

Responsibilities

• Develop and validate machine-learning models, training pipelines, inference systems, and supporting infrastructure.
• Implement model components, data pipelines, evaluation systems, and numerical methods.
• Build reproducible programmatic workflows using Python and command-line tools.
• Work with tensor operations, automatic differentiation, model architectures, tokenization, batching, and generation.
• Optimize training and inference systems for latency, throughput, memory usage, and hardware utilization.
• Diagnose numerical instability, incorrect tensor behavior, memory bottlenecks, distributed-system failures, and performance regressions.
• Compare model implementations and verify that results are correct, reproducible, and performant.
• Review AI-generated code and technical solutions for correctness, efficiency, and engineering quality.
• Design objective tests, benchmarks, and verification criteria for ML systems.
• Document technical decisions, trade-offs, implementation details, and limitations clearly.

Required Qualifications

• Master’s degree or PhD in Computer Science, Machine Learning, Artificial Intelligence, Applied Mathematics, Statistics, Engineering, or a closely related quantitative discipline.
• Strong professional or research experience in machine learning.
• Practical proficiency in Python.
• Meaningful experience with at least two relevant ML frameworks, libraries, or inference tools.
• Strong understanding of model training, evaluation, numerical computation, or inference.
• Ability to debug machine-learning systems beyond surface-level API usage.
• Ability to clearly explain implementation decisions, performance trade-offs, and system failure modes.
• Experience building reproducible technical workflows.
• Demonstrated ability to work with complex ML engineering problems independently.

Relevant Technologies

Experience may include:

• PyTorch
• JAX
• NumPy and SciPy
• SGLang
• vLLM
• llama.cpp
• Hugging Face Transformers
• Hugging Face Tokenizers
• Equivalent tools demonstrating directly relevant technical depth