Data Scientist, Behavioral Science & AI Evaluation

Company: STAR FIND AI INC.

Product: Sayglade™

Location: Kitchener/Waterloo, Ontario, Canada

Work arrangement: Hybrid

Employment: Full-time, 37.5 hours per week

Annual base salary: CAD $92,000–$100,000

Sayglade™ helps people understand everyday conversations and prepare what to say next. Through Decode My Chat and Rehearse My Talk, users can explore possible interpretations of a message, consider thoughtful replies, and practise conversations before having them.

We are hiring a Data Scientist to improve the quality and usefulness of these experiences through statistical modeling, natural language processing, and product experimentation.

You will work closely with product and engineering to turn communication challenges into measurable questions, develop practical analytical methods, and use evidence to guide product improvements. This is a hands-on role in an early-stage product serving people navigating friendships, relationships, university life, and the transition into work.

In this role, you will:

• Prepare and analyze data: Use Python, R, and SQL to extract, clean, organize, and analyze structured product events, user feedback, and authorized conversation evaluation datasets. Develop reproducible workflows and check data quality.

• Develop analytical models: Build, validate, and apply statistical and NLP models to categorize communication scenarios, identify response-quality failure patterns, and examine relationships between product experiences and user behavior.

• Build AI evaluation methods: Develop evaluation datasets, automated scoring methods, and human-review guidelines for chat interpretation, suggested replies, and conversation practice. Assess whether responses distinguish evidence from assumptions, remain relevant to the user’s context, and provide useful next steps.

• Compare models and product approaches: Design controlled evaluations of language models, prompts, and response strategies. Analyze performance across scenarios and document uncertainty, limitations, and trade-offs.

• Measure product outcomes: Analyze activation, repeat usage, and retention. Design experiments and apply appropriate statistical methods to assess product changes, selecting methods suited to the available sample size and data quality.

• Implement validated improvements: Collaborate with engineering to integrate useful analytical methods and evaluation checks into product development, and assess whether changes improve results.

• Communicate findings: Produce clear visualizations, concise research reports, and prioritized recommendations that connect technical findings to product decisions.

All research and evaluation must respect user privacy and data-use permissions. Evaluation datasets may include synthetic, appropriately licensed, or explicitly consented material; private conversations are not automatically available for research or model training.

You should bring:

• A master’s or doctoral degree in statistics, data science, quantitative psychology, psychometrics, educational measurement, or a related quantitative discipline.

• Applied experience in statistical modeling or machine learning, including selecting methods, validating results, and explaining limitations.

• Strong programming and data-analysis skills in Python, R, and SQL. Proficiency in all three is required.

• Experience working with behavioral, survey, product, or text data.

• A solid understanding of experimental design, statistical inference, and measurement reliability and validity.

• The ability to translate an open-ended business or research question into a practical analysis and communicate findings clearly in English.

• The ability to work independently and collaborate with product and engineering colleagues.

Experience with NLP, generative AI evaluation, causal inference, longitudinal analysis, or consumer product analytics would be an asset. We welcome candidates whose experience combines academic research and industry applications.

You will have the opportunity to establish how Sayglade measures AI quality, identifies meaningful product improvements, and applies behavioral science to everyday communication.

To apply, submit your résumé and a brief description of a relevant research or analytics project. Explain the question you addressed, your contribution, the methods you used, and how the findings informed a decision. Please share only material you are authorized to disclose.

Learn more at https://sayglade.com.