Kelp Forest Grove · Work

Work

Where I have worked, and what I actually built there.

Timeline

2024 – present
Staff Data Scientist @ GrowTherapy
I build the data, evaluation, and experimentation systems that make AI and marketplace products measurable, trustworthy, and useful in production.
- AI systems: structured LLM extraction, MCP semantic-layer integrations, LLM evals, AI workflow tooling.
- Measurement: Statsig Warehouse-Native migration, causal inference, rollout measurement.
- Marketplace ML: ranking, forecasting, provider quality, supply-demand, LTV, price elasticity.
2019 – 2024
Senior Data Scientist, Pod Tech Lead @ DoorDash
Led the overhaul of merchant telemetry and schemas. Drove cross-team app quality and measurement QA. Ran A/B and diff-in-diff experiments. Partnered on the experimentation platform and council. Built segmentation and live-cluster pipelines. Owned Snowflake/Databricks/Airflow ETL and the Segment source-of-truth. Shipped funnels and dashboards in Sigma/Tableau/Mode.
Pre-2019
Early Career @ Finance & Analytics
Tufts Economics → Northwestern MS Analytics. Early finance and analytics roles across London and Shanghai. Military leadership in Singapore.

Case studies

1.
Provider Intelligence Platform
structured LLM extraction + identity resolution + Snowflake/dbt + Prefect.
2.
LLM Evals for Analytics Outputs
quality, latency, structured-output validation, model-iteration checks.
3.
Semantic Layer → Claude via MCP
governed natural-language analytics.
4.
Experimentation Platform Migration
Statsig WHN, identity resolution, validation, causal measurement.
5.
Marketplace Intelligence
ranking, forecasting, supply-demand, provider quality.

Merchant telemetry, experimentation platform, and live-cluster pipelines

Economics → Analytics → Data Science