Evan – Dagster, Snowflake, FastAPI

Own the production data platform behind marketplace APIs end to end: multi-source Python ingestion into an Apache Iceberg/S3 lakehouse orchestrated by Dagster, incremental dbt transformations, real-time event streaming on Redpanda (Kafka), and a Postgres + FastAPI serving layer.

• Owned and operationalized the production data path behind marketplace APIs, normalizing multi-source Python ingestion into an Apache Iceberg/S3 lakehouse with Dagster orchestration.
• Designed and implemented real-time event streaming on Redpanda (Kafka-compatible), moving the platform from batch-only to streaming ingestion.
• Led the migration of API serving from Iceberg/Parquet files to Postgres for low-latency reads, keeping Iceberg as the analytical source of truth.
• Rebuilt the transformation layer with incremental dbt models, enforcing canonical grains, column-level lineage, schema guards and parity tests.
• Built statistical data-quality controls for pricing data using bimodal-distribution detection.
• Built a production AI-assisted analytics interface (Cursor Agent SDK) with guarded SQL execution and live-source fetchers.

Owned the Sui DeFi data platform end to end as the primary data partner to the Heads of Partnerships, DeFi and Treasury.

• Designed and shipped the Sui DeFi data platform from zero: reverse-engineered on-chain schemas into dbt models orchestrated by Dagster on Snowflake, powering the KPIs for a $75M ecosystem incentives program.
• Set the metric definitions (retention, growth, stablecoin adoption) that leadership used to steer incentive spend.
• Architected governed LLM access to Snowflake via an MCP layer with SQL allowlists, row caps and caching, making production data safely queryable by AI agents.
• Extended external coverage by integrating financial APIs into the warehouse through Dagster.
• Took initiative upstream: shipped four Rust PRs to the DeepBook v3 indexer (event handlers, Postgres migrations, a portfolio read API) to fix data at the source.

Sole builder of the protocol's analytics platform, from warehouse design through published research.

• Designed and built the analytics platform from scratch: Python ETL ingesting consensus, mempool and protocol event data into a self-managed Postgres/TimescaleDB warehouse, served via PostgREST and materialized views.
• Turned transaction-level data into research on inclusion latency, censorship incentives and rollup reliability; the series won an Ethereum Foundation grant.
• Developed an execution-quality measurement methodology, published in Frontier.

Customer Data Analytics at Dune Analytics (Fall 2021) — partnered with engineering on the technical roadmap from customer feature requests and feedback.

Partnered with engineering on the technical roadmap from customer feature requests and feedback.