Staff Data Scientist in at Nerdy

Overview:

As the Staff Data Scientist for Conversation Intelligence, you will apply natural language processing and large language models to the conversations at the heart of Nerdy's business. You will design the methods that turn unstructured speech and text into reliable, measurable signals, build and rigorously evaluate LLM-based systems, and take models from research to production on a modern lakehouse stack. You will own your problem area end to end and be accountable for the business outcomes your work moves, leveraging AI to accelerate development, evaluation, and operations.

About Nerdy:

At Nerdy (NYSE: NRDY) - the company behind Varsity Tutors - we’re redrawing the blueprint of learning. Our Live + AI™ platform fuses real-time human expertise with proprietary generative-AI systems, setting a new bar for measurable academic impact at global scale.

We recruit the kind of technologists and operators you’d bet on as solo founders - people who turn ambiguous problems into shipping code, iterate faster than markets move, and compound their advantage with every data point. In an era where great employees can deliver 10-times the leverage of the merely good, we back those who play to win.

Fortune favors the bold. Join us.

How we compete:

• AI-Native at every level From the CEO to day-one hires, everyone builds and ships with generative AI. If you’re not wielding AI, you’re not done.
• Entrepreneurial velocity Move at founder speed, prototype in hours, and measure in real user outcomes. Slow teams die.
• Free-market rigor Ideas rise or fall on merit and results - no committees, no politics, no cap on upside.
• Full-stack ownership You design, build, and run what you ship; accountability is a feature, not a bug.
• Reward for contribution Pay rises with impact, not years. Outstanding results earn outsized rewards. We evaluate both what you achieve and how you achieve it: living our leadership principles and using AI effectively are formally measured and rewarded.
• Relentless exploration Push the frontier of generative AI in live learning and - because only the paranoid survive - questioning every legacy assumption along the way.
• Is Apolitical You stay focused on mission-aligned outcomes, not distractions or unrelated causes.

If you’re a technically minded builder who thrives on open competition, personal responsibility, and the chance to redefine how the world learns - while continually stretching the limits of what generative AI can do - come do the most ambitious and rewarding work of your career here. Learn more at nerdy.com.

Nerdy’s shareholder letters below explain our latest products and strategy:

• Q2-2026 Shareholder Letter
• Q1-2026 Shareholder Letter
• Q4-2025 Shareholder Letter

Qualifications:

• Master’s degree in Computer Science, Statistics, Mathematics, Computational Linguistics, or a related quantitative field is required; a PhD is preferred.
• 6+ years applying data science or machine learning to real products, including at least one system you took from concept to production and kept in use.
• Expertise in natural language processing and large language models: prompt design, structured extraction, fine-tuning or retrieval where it pays off, and the judgment to know when a simpler model wins.
• Hands-on experience evaluating LLM systems: reference datasets, rubric design, model-assisted review calibrated against human reviewers, inter-rater agreement, and regression testing whenever prompts or models change.
• Strong Python (pandas, scikit-learn, PyTorch or similar) and expert SQL on a modern warehouse or lakehouse. We run Starburst Galaxy (Trino) on Iceberg with dbt Cloud; Snowflake, BigQuery, Databricks, or Athena experience transfers.
• Experience with the major LLM APIs (Anthropic Claude, OpenAI, or similar) and the tooling around them, including cost, latency, and privacy constraints at production volume.
• Comfortable with speech-to-text output and messy human-generated text: speaker attribution, long-context summarization, PII handling, and the errors transcription introduces.
• Experimentation and causal inference skills sufficient to connect a model-driven change to a business outcome, not just a model metric.
• Ability to explain findings to non-technical business leaders in their terms, and to turn their questions into measurable definitions.
• Git-based development, code review, and CI/CD habits; you write code other people can run and maintain.
• Experienced with AI-native tools that enhance productivity and speed (e.g., Cursor, Make, Supabase, Netlify, Claude Code, n8n, Firecrawl, ChatGPT, Grok, Bolt, Vercel, etc).
• Alignment with Nerdy’s apolitical, mission-focused culture.

Responsibilities:

• Conversation modeling: Apply NLP and large language models to call, chat, and session transcripts to extract structured, reliable signals from unstructured speech and text.
• Measurement design: Work with business partners to define what good looks like in a conversation, and translate those definitions into validated, measurable metrics.
• LLM evaluation: Build the evaluation methods, including reference datasets, rubrics, and model-assisted and human review, that establish whether an LLM-based system is accurate, consistent, and safe to rely on, and that catch regressions when models or prompts change.
• Modeling and analysis: Build classification, extraction, scoring, and summarization models, and the analyses that connect conversation behavior to business outcomes such as conversion, resolution, and retention.
• Production: Take models from prototype to production with engineering partners, with monitoring and retraining so they stay reliable in use.
• Experimentation: Design and analyze experiments that attribute business impact to changes informed by your models.
• Data foundations: Partner with Data Engineering and Analytics Engineering to make transcripts, labels, and model outputs available in the lakehouse (Starburst, dbt, Supabase) as governed, reusable tables.
• Communication: Present findings and recommendations to business and technical leaders in their language, with the evidence behind them.
• AI-enabled workflows: Use AI for code generation and review, experiment analysis, transcript exploration, and documentation, operating within privacy and security guardrails.Overview

As the staff data scientist, you will operate with high independence to discover insights and build production ML systems that personalize live learning at scale. Based in South America, you’ll partner with Product and Engineering to take ambiguous problem spaces from definition to deployed models—spanning LLM-driven summarization, real-time ranking, recommendations, and churn prediction.

Responsibilities

• Own end-to-end delivery of customer-facing ML systems: problem framing, qualitative analysis, data prep, modeling, validation, deployment, and monitoring.
• Apply and extend generative AI/LLM techniques (e.g., RAG, agents, summarization, content generation, personalization) to real-world learning experiences.
• Define and run offline/online evaluation: success metrics, A/B tests, counterfactuals, model QA, drift detection, and continuous improvement loops
• Partner with Product and Engineering to scope, prototype, and productionize solutions that meet latency, reliability, and privacy requirements.
• Drive data strategy for ML: feature definition, labeling/ground truth, instrumentation, and data quality SLAs; contribute to feature store and streaming feeds.
• Build and maintain robust MLOps/LLMOps practices: model/prompt versioning, CI/CD for models, canaries/shadow deploys, observability, and rollback strategies.
• Communicate insights and trade-offs clearly to technical and non-technical audiences; influence roadmaps with data and experimentation.
• Mentor peers across Data Science/Analytics; review work, share best practices, and raise the technical bar for the org.

Qualifications

• M.S./Ph.D. in a quantitative field (e.g., Mathematics, Statistics, Machine Learning, Econometrics) or equivalent experience.
• 8+ years of professional data science experience, including shipping production ML in customer-facing products.
• Deep expertise in statistical inference, probability, and supervised/unsupervised learning; strong product intuition and bias for action.
• Proficiency in Python and SQL; experience with ML frameworks (e.g., scikit-learn, XGBoost, PyTorch/TensorFlow) and experiment platforms.
• Hands-on experience with LLMs/gen AI (RAG, agents, prompt engineering/evaluation) alongside classical ML (classification, ranking, recommendation).
• Experience with data and platforms: streaming (Kafka/Kinesis), warehousing (Redshift/Snowflake), orchestration (Airflow), and feature stores.
• Strong MLOps fundamentals: model serving, monitoring, drift/quality management, and reproducibility.
• Excellent communication and stakeholder management; proven ability to lead ambiguous work across functions.
• Experienced with AI-native tools that enhance productivity and speed (e.g., Cursor, Make, Supabase, Netlify, Claude Code, n8n, Firecrawl, ChatGPT, Grok, Bolt, Vercel, etc).
• Understand and appreciate that Nerdy is an apolitical company focused on helping people learn.