Two engagements with the company that defines behavioral data. As Solutions Architect, I led data projects for large enterprise customers on AWS and GCP: architecture diagrams, infrastructure managed with Terraform, custom dbt models, and behavioral pipelines loading BigQuery, Snowflake, Redshift and Databricks. As Principal Solutions Engineer, I designed and built Snowplow's central field demo platform: seven complete industry applications with live Signals personalization and AI concierge agents, which the global sales organization now runs in front of Fortune 500 buyers.
Large companies don't struggle to install Snowplow. They struggle to fit it into their cloud, their warehouses, their privacy obligations and their existing pipelines. That was my job: turn a powerful tool into a working enterprise architecture, with diagrams their engineers could actually build from.
I led data projects for large enterprise customers end to end, using Terraform to manage Snowplow infrastructure as code and landing behavioral data in BigQuery, Snowflake, Redshift and Databricks. One architecture per client, designed around their cloud and their constraints rather than a reference diagram.
Client dbt models built for BigQuery, Snowflake, Databricks and Redshift, turning raw behavioral events into the entities their analysts actually query: sessions, users, funnels and marketing attribution.
I designed data protection strategies for behavioral flows from web applications, including user and session cookie management, consent handling and data minimization. Built to pass enterprise compliance review.
Cloud infrastructure and ELT pipelines architected with Kafka, Airflow, Pub/Sub and Kinesis, matched to each client's stack. The deliverable was infrastructure their teams still run.
Enterprise buyers do not believe slides. Snowplow's field team needed real applications with real tracking, behavioral attributes computed in seconds, and AI agents that act at the moment of intent, for every industry they sell into. I built the whole platform, from first commit to a finished product: media, ecommerce, fintech, gaming, SaaS, jobs and travel.
A monorepo of seven industry applications on shared packages, with a tracking design for each industry, custom schemas in development and production registries, provisioning scripts, dbt models and demo talk tracks. I designed and shipped the shared platform.
Live behavioral attributes and published interventions for every industry: abandoned booking recovery, concierge offers triggered by comparisons, personalized paywalls and campaign prompts. Every intervention fires from live Signals state. What the buyer sees is the product reacting to their own behavior, not a script.
Signals proactively opens AI concierge agents, for example when a traveller compares destinations without booking. Agents run through the Vercel AI Gateway, grounded strictly in local product catalogues so they never hallucinate in front of a buyer, with scoring fallbacks when the model is unavailable.
Presenter panels expose live attributes and intervention status, demo control bars handle session resets and UTM presets, and consent defaults plus intervention fallbacks keep the demo alive end to end. Account executives can run the demos without an engineer in the room.
Every vertical continues past the app into the data, with Snowflake Notebooks for all seven demos plus BigQuery analytics notebooks: raw event exploration, dbt behavioral modeling, and the warehouse narrative that closes technical evaluators.