Data Scientist, Snapdocs

• 3+ years in data science, preferably working on ML-driven products
• Experience with a wide range of machine learning approaches, including LLMs and traditional supervised/unsupervised learning
• Background in document classification or information extraction (text and images)
• Experience collaborating with Product and Engineering to ship models into production
• Expert in Python and ML libraries (e.g., Scikit-learn, HuggingFace)
• Comfortable with SQL and working in production data pipelines
• Hands-on experience with cloud platforms (AWS, GCP, or Azure)
• Understanding of service architecture and working with APIs
• Familiar with CI/CD practices and software engineering fundamentals
• A collaborative, thoughtful problem solver
• Curious and passionate about ML, NLP, and Generative AI
• Comfortable working on ambiguous problems with imperfect data
• Driven to turn research into scalable, production-ready systems

Desirable

• Snapdocs is looking for a Data Scientist to help us improve and scale our Document Quality Control (QC) services. You’ll partner with our lead data scientists to develop and optimize solutions for document classification, information extraction, and annotation detection—across hundreds of document types
• This is a high-impact role where you’ll apply the latest in NLP and Generative AI to real-world use cases, helping us build reliable, intelligent systems that make sense of messy, unstructured document data
• Improve the performance and generalizability of Document QC services, including classification, extraction, and annotation detection
• Apply and test cutting-edge generative AI methods to improve outcomes across varied document types
• Optimize model performance for new customer document sets
• Partner with DS Operations to ensure high-quality training and evaluation data
• Contribute to service repositories and help productionize models in collaboration with Engineering
• Your first project will be either:
• Testing, implementing, and optimizing a general classification methodology on a new customer dataset
• Or
• Deploying and refining an information extraction pipeline for a new customer use case