Description

Position Type: Contract/Affiliate Resource Pool Duration: Immediate need: Remote, 20-40 hours over 6 weeks. Thereafter, project-based engagements (2-6 months typical) Location: Remote/Hybrid (travel to client sites as needed)

We are seeking a highly skilled Multimodal AI Training Data Scientist to join our elite resource pool of AI transformation specialists designing and implementing cutting-edge multimodal AI solutions. The ideal candidate will drive enterprise AI initiatives that solve complex technical document understanding problems and transform how organizations interpret critical engineering and industrial drawings.

Key Responsibilities

Core Requirements:

Design and implement sophisticated training data creation workflows for technical document understanding
Collaborate closely with process engineers and domain experts to rapidly acquire technical knowledge and ensure accuracy
Develop comprehensive annotation strategies and quality control frameworks for engineering and industrial applications
Create multimodal training datasets that enable AI systems to interpret complex technical symbols, equipment specifications, and process workflows
Work with subject matter experts to validate technical accuracy and industry compliance of training data
Support high-stakes model deployment activities and post-implementation performance optimization

Must-Have Qualifications

Ideal Candidate:

Multimodal AI & ML Foundations. 5+ years architecting and deploying vision–language models (VLMs) for technical document analysis in production, with deep expertise in computer vision, NLP, and multimodal AI system design.
Fine-Tuning & Model Optimization. Proven track record fine-tuning large language models (e.g., Google Gemini, GPT-4V, Claude) and implementing advanced computer vision techniques such as object detection, symbol recognition, and document analysis for complex technical drawings and diagrams.
Training Data Engineering. Expert in designing instruction-following datasets, conversation formats, and robust quality assurance frameworks to ensure technical accuracy, aligned to business requirements.
Document AI Solutions at Scale. Enterprise-level experience developing PDF processing pipelines with OCR, image analysis, annotation workflows, structured data extraction, and ML training infrastructure that deliver measurable business value.
Programming & Deployment. Python mastery with ML libraries (PyTorch, TensorFlow, OpenCV, transformers), cloud platforms, and production-grade AI deployment using Agile methodologies.
Collaboration & Communication. Ability to collaborate with subject matter experts to rapidly acquire domain knowledge in technical or regulated industries with strong written and verbal skills.
Eligibility. Canadian or U.S. citizen, or visa holder authorized to work in the U.S. (no sponsorship available).

Experience & Education

Work Experience

5+ years in Multimodal AI & Machine Learning Background in Engineering, Industrial processes, or technical document analysis (preferred but not required)

Education

Master's degree in Data Science, Computer Science, Machine Learning, or related field (or equivalent experience)

About Flatirons Digital Innovations

Flatirons Digital Innovations Inc., FDI, is a digital transformation agency that secures data, safely moves data, and liberates historical information to create new experiences and generate new revenue opportunities. We solve discrete business and customer experience challenges through content and data lifecycle services, the development of custom solutions through our Innovation Studio, and by unlocking data with our Flatirons Digital Hub platform-as-a-service.

With roots in Boulder, Colorado, our team of inspired transformation experts have solved complex challenges for over 20 years.

Flatirons Digital Innovations

Latest Jobs

Data Scientist (Remote)