Remote job
Research Engineer: Inference / MLOps
Job details
About this role
Role overview Take machine learning models from working checkpoints to reliable, performant production serving. This role combines inference engineering, deployment tooling, and operational ownership, with optimization choices tailored to different model needs—for example, low-latency streaming versus slower, quality-sensitive dubbing.
Responsibilities - Deploy models to production and own the path from checkpoint to serving infrastructure. - Improve output quality through approaches such as guidance, guardrails, and automated detection of errors or hallucinations. - Reduce inference latency and cost, and increase throughput using techniques such as kernel, cache, GPU scheduling, or quantization improvements. - Build monitoring and incident flagging for inference failures, timing anomalies, and model-level statistics. - Create deployment tools that move new models rapidly from development to scaled serving while making effective use of GPU capacity. - Make independent optimization decisions based on the requirements of each model and use case.
Benefits and work setup - Professional development and career growth opportunities. - Competitive compensation, performance bonuses, and comprehensive benefits.