Document AI Engineer: ML Ops & OCR

boxinc; United States; Hybrid; USD 120,000 - 170,000; Full time

Benefits offered by this job

Hybrid work schedule

Job summary

boxinc is hiring to build and operate machine learning systems that turn complex enterprise documents into structured, usable information. You will work across document understanding, including OCR and language-model-based extraction, from development to production with cross-functional teams.

The role emphasizes strong Python skills, collaboration, and ownership while supporting real-time and batch workloads.

Qualifications

• At least 3 years of experience building, deploying, or operating production software, data, or ML systems.
• Strong Python skills, with readable, testable production code.
• Experience with the lifecycle of language-model-based agents, including data prep and prompting.
• Clear communication, collaboration, and ownership mindset for ambiguous problems.

Responsibilities

• Develop, deploy, and maintain ML models, language model agents, and supporting services.
• Build data prep pipelines and prompts, tools, and agent workflows for document extraction.
• Measure model quality with evaluation sets, experiments, metrics, and feedback.
• Monitor production systems for reliability, latency, cost, and data/model changes; participate in on-call.
• Design scalable services for real-time and batch workloads; provide tests, docs, and runbooks.
• Collaborate with partner teams to translate customer needs into production capabilities.

Skills

Python

Communication

Collaboration

Ownership mindset

Tools

Kubernetes

Terraform

SQL

Job description

boxinc is hiring to build and operate machine learning systems that turn complex enterprise documents into structured, usable information. You will work across document understanding, including OCR and language-model-based extraction, from development to production with cross-functional teams.

The role emphasizes strong Python skills, collaboration, and ownership while supporting real-time and batch workloads.