Note: The job is a remote job and is open to candidates in USA. Candid is a nonprofit that provides data and insights about the social sector. The Data Scientist will apply machine learning and statistical methods to data quality, model evaluation, prototyping, semantic search, and other data-science initiatives while collaborating with product, data, engineering, analytics, and subject-matter teams.

Responsibilities

• Apply ML across data problems: detection, classification, entity resolution and matching evaluation, anomaly detection, embeddings and semantic search, and feature engineering
• Measure how models perform on real production data (not just test sets), build ground-truth where labels don’t exist, and detect drift and degradation
• Apply ML to data quality: find systematic, at-scale errors that queries and manual review miss, and route findings to the people who own the fixes
• Take open-ended questions through to a prototype and a clear recommendation, including ruling ideas out when they aren’t worth building
• Build enabling tooling and dashboards (for example, semantic search over text, or Streamlit dashboards) that help analysts and stewards do their work
• Work embedded with a product or data team on project-defined scope, partnering with engineers, analysts, PMs, and subject-matter experts
• Take on other ML and data-science projects as Candid’s needs and priorities shift

Skills

• Applicants for this position must be legally authorized to work in the United States without employer sponsorship, now or in the future
• 3-5 years of relevant experience
• A strong generalist applied data scientist, comfortable moving across classification, entity-resolution and matching evaluation, anomaly detection, embeddings and semantic search, feature engineering, and exploratory feasibility work
• Strong Python and SQL, comfortable with large production datasets (Starburst/Trino, Snowflake) and working in AWS
• Experience measuring model performance on real-world / production data (not just test sets), including building ground-truth where labels don't exist, and detecting drift
• Comfort building and maintaining dashboards (the team's are in Streamlit) that surface data-quality and model-performance metrics
• Experience taking a fuzzy question to a prototype and a recommendation, and being willing to rule an idea out
• Sound statistical judgment: sampling, error rates, uncertainty, and knowing when a finding is real
• Strong communication and comfort working embedded in another team with project-defined scope
• Sensitivity and respect for racial, gender, sexual orientation, and cultural differences
• Preferred someone with experience applying ML to data quality, such as detection or classification models that flag anomalous or wrong records at scale
• Preferred (any of these are a plus): embeddings, semantic search, or NLP over text corpora; recommendation, ranking, or propensity modeling; computer vision or multimodal ML for image-based QA; experience evaluating entity-resolution or deduplication systems; turning exploration into derived-data products; and familiarity with the U.S. nonprofit and philanthropic sector

Benefits

• Health insurance (medical, dental, vision)
• Retirement contribution with additional option for a match
• Paid life insurance and AD&D
• Paid leave time (PTO, compassionate leave, volunteer, holiday, parental)
• Short-term and long-term disability
• Pre-tax transit
• Flexible spending accounts
• Supplemental insurance
• Summer hours
• Public Service Loan Forgiveness (PSLF) program eligible employer
• Remote work arrangement
• In-person attendance is expected twice per year during our annual, weeklong all-staff summits

Company Overview