Freelance data scientist and analytics engineer
I am Syed Ahmed Basharat Ali, a freelance data scientist and analytics engineer based in New York City, working remotely with companies anywhere. Through my own consulting practice I was Lead Data Scientist at a Saudi edge-AI startup, where I found and fixed an 80% silent data loss, stabilized the database behind it and moved the warehouse onto dbt. Before consulting I was a data scientist at Geidea, Saudi Arabia's largest payments processor, and a senior data consultant at EY.
I take on fixed-scope projects and ongoing part-time work. Most engagements start with a short call about the problem and a written scope.
What I can do for you
- Analytics engineering and dbt migrations: Moving warehouse logic onto dbt, modeling multi-tenant Snowflake and PostgreSQL data, and defining metrics so the same question returns the same number.
- Data reliability, reconciliation and observability: Finding silent data loss, stabilizing failing syncs, and building reconciliation and monitoring that catch failures before they reach a dashboard.
- Payments, fraud and settlement analytics: Anomaly detection, fraud and risk models, settlement and reconciliation data, and merchant churn and revenue forecasting for payments businesses.
- AI assistants and LLM evaluation: Building AI assistants on OpenAI and Anthropic APIs, and measuring whether LLM outputs, and the numbers reported about them, can be trusted.
- Fractional head of data: Acting as the data lead for an early-stage company: owning the warehouse, the metrics and the reporting the CEO uses to run the business.
Client work through my consulting practice
- Saudi edge-AI startup (client): traced an 80% upstream data loss across a 150-device camera fleet and shipped a staged fix.
- Saudi edge-AI startup (client): stabilized a 137 GB PostgreSQL backbone that was failing 8-hour Airbyte syncs.
- Saudi edge-AI startup (client): shipped an 11-model dbt migration across a multi-tenant Snowflake warehouse (50+ client schemas), with reconciliation and observability built from zero; it surfaced two real sync issues on its first run.
- A KSA grocery-delivery startup (client): a bilingual Arabic/English AI assistant on OpenAI function calling, and CatBoost stock-availability prediction.
Earlier work
- Geidea (Aug 2023 to Jan 2025): real-time anomaly detection on 15M+ daily payment transactions that evolved into production fraud models; churn and revenue forecasting behind a retention strategy that cut merchant churn from 35% to about 10%; collections automation contributing to $8M+ in additional annual profit.
- EY (Jan 2022 to Jun 2023): founding member of the EY Pakistan Business Consulting data team; C-suite dashboards for GSK and National Foods.
Public work you can inspect
- Silent-failure detection for a distributed capture fleet: Row counts catch 18 of 43 failures, content checks catch 43 of 43; bad device-days fall from 1,097 to 58.
- A trust harness for a semantic layer: 40 of 90 queries (44%) have no single answer; one revenue question returns $149M, $136M or $107M depending on the join.
- Validating the LLM judge that decides what you bill for: The LLM judge overstates resolution by 5.2 points overall and +19.4 in one category; the judge-only confidence interval covers the truth 0% of the time.
Tools
Python, SQL, dbt, Snowflake, PostgreSQL, Airflow, Airbyte, Tableau, Power BI, OpenAI and Anthropic APIs, AWS, Docker, GitHub Actions.
Contact
Email sabasharat.ali@gmail.com, or find me on LinkedIn and GitHub. Full history on the experience page.
Available for freelance and contract work, remote or in New York.