Ankit B.

Marketing says it sourced $4M in pipeline. Sales says $1.2M. Finance has a third number. Nobody can explain the gap. That's what I fix — the data layer underneath B2B marketing and revenue reporting. I'm a data engineer with 10+ years building marketing and revenue data infrastructure, including five years doing exactly this at Expedia and Meta. At Expedia I built the marketing data warehouse and productionized first-touch, last-touch and multitouch attribution models into Salesforce campaign performance reporting. At Meta I worked on ad spend attribution, advertiser billing reports, and clicked/exposed booking reporting — environments where the number had to be defensible, because money moved on it. WHAT I DO - Attribution and pipeline reconciliation — reconcile CRM opportunity revenue against ad platform spend, marketing automation and web analytics; find where the numbers diverge and why - Marketing data warehouse builds — Fivetran and Airflow ingestion from Salesforce, ad platforms and product analytics into Snowflake or BigQuery, modeled in dbt - Attribution models in production — first/last touch and multitouch, built to survive late-arriving conversions, backfills and schema drift, not sitting in a notebook - Campaign taxonomy and mapping governance — campaign ID to internal taxonomy, UTM standards, the unglamorous layer that silently breaks every number downstream of it - Revenue and pipeline reporting — Looker, Looker Studio, Tableau and Power BI on a governed semantic layer, with metric definitions your team actually agrees on STACK SQL, Python · Snowflake, BigQuery, Databricks, Redshift · dbt, Airflow, Fivetran · Salesforce, Google Analytics 4 · Looker, Looker Studio, Tableau, Power BI · AWS, GCP · Git, Terraform, Docker Certified: AWS Certified Data Engineer – Associate · SnowPro Advanced: Data Engineer · dbt DATA FIRST, THEN AI Most AI initiatives don't fail on the model. They fail on what's underneath — metric definitions nobody agreed on, broken lineage, pipelines the business already doesn't trust. Layering AI on that just produces confident answers from bad inputs. So I work in that order. Get the data layer right — governed, documented, reconciled — then the AI work has something solid to sit on. If someone has told you your data isn't AI-ready, I'll tell you specifically what's missing, what it costs to fix, and what the first AI use case should be once it is. HOW I WORK Seattle-based, US hours. Everything I build is documented and owned by you — no proprietary implementations, no black boxes, nothing you need me to maintain. If your campaign reporting doesn't reconcile with your CRM, or your board deck numbers change every time someone fixes a mapping, send me the details and I'll tell you what I'd look at first.