ML Engineer: Research-to-Production Pipelines

Deepgram; United States; On-site; USD 180,000 - 240,000; Full time

Job summary

Deepgram is seeking an Applied ML Engineer to own the research‑to‑production pipeline and streamline model delivery at scale. You’ll work with researchers to productionize novel models, build robust tooling, and ensure repeatable, observable deployments.

You’ll be responsible for designing release gates, optimizing inference, and delivering scalable ML systems across our hybrid infrastructure.

Qualifications

• Must have production‑quality ML code in Python.
• Experience shipping ML models from research to production at scale.
• Strong knowledge of PyTorch and the modern DL stack.
• Experience building ML pipelines and developer tooling (training orchestration, evaluation harnesses, packaging, deployment, or CI/CD).
• Familiarity with inference optimization (latency, batching, throughput) for production.
• Comfort operating across distributed systems and GPU compute (cloud and on‑premise).

Responsibilities

• Own the research‑to‑production pipeline: turn research checkpoints into production models and define repeatable deployment paths.
• Collaborate with researchers to productionize new models with robust, reproducible workflows.
• Build tooling and abstractions to move models through training, evaluation, packaging, and deployment with minimal friction.
• Design and own model release gates with automated evaluation, regression checks, and timing decisions.
• Optimize models and serving for production: reduce latency, improve throughput, and enable cost‑efficient inference.
• Strengthen the build and delivery layer for models on hybrid GPU/cloud infrastructure.
• Establish benchmarking and validation that catch regressions from development to production.
• Create a feedback loop to surface model behavior and accelerate next iterations with research.

Skills

Python

Distributed systems

PyTorch

ML pipelines

Inference optimization

Production ML

Research-to-prod

Tools

PyTorch

Job description

Deepgram is seeking an Applied ML Engineer to own the research‑to‑production pipeline and streamline model delivery at scale. You’ll work with researchers to productionize novel models, build robust tooling, and ensure repeatable, observable deployments.

You’ll be responsible for designing release gates, optimizing inference, and delivering scalable ML systems across our hybrid infrastructure.