Описание:
FieldFlo is a mobile-first SaaS platform serving the construction, demolition, and environmental services industries. It supports compliance, safety, time tracking, training, and field operations for teams across the U.S.
Задачи:
- Design and deploy document understanding pipelines for PDFs, drawings, forms, and other unstructured documents
- Train and optimize models such as Donut, LayoutLM, and TrOCR, or similar document AI architectures
- Build multi-stage AI workflows combining OCR, computer vision, and LLM-based reasoning
- Develop object detection and extraction models to identify regions of interest, annotations, tables, labels, and document elements
- Convert extracted information into structured outputs with confidence scoring and traceability
- Build evaluation frameworks, monitoring, and retraining workflows
- Collaborate with engineering and product teams to ship AI features into production
Требования:
- 3+ Years of professional ML Engineering experience
- Production experience with Document Understanding solutions
- Hands‑on experience training or deploying models such as Donut, LayoutLM, TrOCR, OCR/document extraction architectures, or equivalent multimodal document AI models
- Experience building end-to-end document processing pipelines, including OCR, information extraction, document classification, structured data extraction, and LLM-based document workflows
- Strong Python skills and experience with PyTorch, Transformers, OpenCV, and the Hugging Face ecosystem
- Experience with AWS, Azure, or GCP Docker and containerized deployment experienceStrong English communication skills
- Будет плюсом: experience with architectural drawings, engineering plans, construction documents, or technical schematics; annotation tools such as CVAT, Label Studio, and Roboflow; agentic AI or multi-step LLM pipelines; PDF rendering, OCR engines, coordinate systems, and document processing frameworks; startup or SaaS experience
Условия:
- B2B contract
- Paid national holidays based on the candidate’s country
- Optional unpaid personal time off
- Three-month probation
- At least 2 hours of overlap with US Mountain Time