Data Scientist

Goodie AI

You ll do:

• Work with large datasets. Own efficient querying, cleaning, labeling, and taxonomy alignment for brands, SKUs, and categories.

• Design sampling and classification strategies that turn noisy LLM outputs and crawler logs into reliable brand and product insights.

• Use LLMs and NLP to extract structure from unstructured text at scale. Topics include query fan-out, sentiment, citation extraction, and entity linking for brands, products, and creators.

• Define product-grade metrics. Create durable definitions for visibility score, answer coverage, product presence, and agentic checkout readiness.

• Build and run experimentation frameworks. A/B tests, holdouts, counterfactuals, and uplift modeling to quantify impact on citations, share of voice, and conversions.

• Develop and refine predictive models that analyze and forecast AI search behavior across models and surfaces.

• Translate complex findings into clear decisions. Partner with the founding team to inform roadmap, pricing, and customer playbooks.

• Create evaluation harnesses. Establish automatic evals and human-in-the-loop labeling for model quality, bias, and drift across LLM providers.

• Detect anomalies. Build monitors for crawler behavior, rankings, and feed health to catch regressions before customers do.

You have:

• 3 to 7 years in applied analytics or data science within tech, marketing, or ads. Startup or high-growth experience preferred.

• Strong Python and SQL. Comfortable in notebooks and in code reviews.

• Skilled with sampling and inference. Stratified sampling, bootstrapping, extrapolation, reweighting, and variance estimation.

• Solid ML toolkit. Time series, classification, regression, weak supervision, and methods to estimate event frequency from partial observations.

• Practical LLM knowledge. Strengths in prompt design, structured extraction, embeddings, and an understanding of model limits and failure modes.

• Curious and current on multi-modal and LLM research. You enjoy reading papers and pressure testing ideas in real data.

• Builder mindset in a fast team. You value clarity, speed, and ownership.

Nice to have:

• Experience with large-scale information extraction or search quality

• Background in causal inference, MMM, or attribution models

• Hands-on work with product feeds and retail catalogs

• Contributions to open source or published work we can read

• Deployed side projects we can click through

Our data and modeling canvas:

• Problems: AI search measurement, AEO scoring, agentic commerce readiness, product catalog and feed integrity, ranking and citation shifts, attribution for AI traffic

• Signals: LLM responses, crawler and agent logs, SERP and AI answer snapshots, product feeds, marketplace metadata, GA4 and GSC connectors, CRM data

• Targets: Share of voice, citation count, answer coverage, SKU presence, conversion lift, time-to-value for optimizations

Tech stack you will touch:

• Languages: Python, SQL

• Libraries: pandas, NumPy, scikit-learn, PyTorch or TensorFlow, Hugging Face, spaCy.

• Data: Postgres or AlloyDB, BigQuery, dbt, DuckDB for local work

• Production ML/MLOps: model serving (FastAPI/Flyte/Batch jobs), CI/CD, versioning, experiment tracking (MLflow/Weights & Biases), monitoring & alerting for performance/drift.

• Cloud & data tooling: AWS/GCP/Azure, containers (Docker).

• Models and providers: OpenAI, Anthropic, Google, Meta, Mistral, Perplexity, together with internal eval harnesses