Document Intelligence and Machine Learning Research Engineer

• Build pipelines that find tables, plots and figures on scanned pages and extract their values with OCR and vision-language models.
• Recover numeric values from plots by reading axes, scales and data points, and record the uncertainty of each value.
• Calibrate model confidence so that a stated confidence matches how often values are right.
• Build benchmarks with gold values checked against the source, and keep their test sets sealed from training.
• Evaluate vision-language models across model families and pinned versions on the same benchmarks.
• Send low-confidence values to human review, and measure how review changes accuracy.
• Link every extracted value to its page and region in the source document.
• Write up methods and results so another researcher can repeat them.

• An MS or PhD in computer science or a related field, or equivalent experience, in document intelligence or computer vision.
• Experience evaluating vision-language models and benchmarks.
• Experience with OCR, layout analysis or table extraction on scanned pages.
• Strong Python and experience training or fine-tuning deep learning models.
• Experience measuring accuracy and calibration on held-out data.

• Experience with PyTorch.
• Experience extracting data from plots and charts.
• Published work on document understanding or chart reading.
• Familiarity with engineering or scientific test reports.

$130,000 to $201,000 a year

Full-time W-2 employment with health insurance, a retirement plan and paid leave.

No application deadline. We accept applications for this role on an ongoing basis.

Remote, United States