Ahmad T.

Your dashboard says one number. Your database says another. Or your new AI assistant answers confidently and gets it wrong often enough that nobody trusts it. Both problems live in the same place, the data layer underneath, and that is the layer I work on. I'm an analytics engineer with 7+ years across fintech, e-commerce and SaaS. I work on the part most people skip: the SQL and dbt models, the grain of the fact tables, the joins that quietly fan out and double your revenue number, and the reconciliation that proves a figure before anyone reports it. HOW I USUALLY START Diagnosis first. I find out which numbers are wrong and why, in writing, with the queries that show it. You get a clear answer on where the breakage actually is before committing to a bigger build. Most engagements then move into fixing it: rebuilding the model, the pipeline, or the reporting layer on top. WHAT I BUILD Data models and warehouses: SQL and dbt on BigQuery, Snowflake, Postgres and Redshift. Grain, tests, documentation, lineage. Pipelines: Python ETL, scheduled loads, API and SFTP ingestion, schema drift handling, validation on ingest. Reporting layers: the metric definitions and semantic layer under Power BI, Tableau, Looker Studio, Metabase and Superset. I build the model underneath, not just the charts on top. Data audits: reconciliation between source systems and reports, GA4 and tracking audits, and the "why do these two numbers disagree" investigations nobody wants to own. Data for AI systems: structured retrieval, text to SQL with guardrails, and the question most AI projects skip, which is whether the warehouse underneath can actually answer what is being asked of it. Plain language querying over a real warehouse is still an open research problem (the best published systems score around 21% on the Spider 2.0 benchmark), so I design for answers you can check against a source, not answers that merely sound confident. TRACK RECORD Payments gateway: lifted online payment success rate 7% month over month, cut bounce rate 35% by instrumenting GA4 and Amplitude properly, then built the SQL, dbt and Looker reporting the sales team ran on daily. Health information exchange: built the Tableau, Looker and Elasticsearch reporting infrastructure behind a 10x rise to 200K+ users. D2C subscription business: owned the analytics for AI personalisation across 50K+ users, moved matching accuracy 12%, cut swap rates 23%, and ran a simulation that saved $10M a year by right sizing portions against calorie targets. Creator marketplace: shipped 15+ Tableau and Redash dashboards, and deployed an internal GPT tool over company data that cut ad hoc analysis response time 60%. CREDENTIALS Top Rated on Upwork, 100% Job Success, 5.0 across every completed contract. BEng in Mechatronics, currently finishing an MSc in Data Science and Artificial Intelligence at the University of Liverpool. IBM Data Analyst and Google Data Analytics Professional Certificates. Tell me which number you don't trust, or what your AI system keeps getting wrong, and I'll tell you where I would look first.