Rhea Anjali Gupta
SHead of Data, Cairn Foods - rules engines · Python · anomaly detection
- Cairn Foods
- San Francisco, CA · PT (UTC-8)
- 14+ years experience
- Northwestern University
- English
Messaging unavailable
Overview
I've spent 18 years on fraud detection - first at PepsiCo, now at Cairn Foods. Taught SQL to 42 non-analysts. Cut false positives on payment blocks by 42%. Most at home with LightGBM and Python. I teach a weekend class on rules engines.
Good reasons to reach out
- Speaking or podcast invitations
Please don't message about
- Data broker offers
- Survey or panel invitations
Slower in the last week of each quarter
Skills
PyTorchSQLLightGBMrules enginesanomaly detectionPython
Experience
- Head of DataSep 2024 - PresentCairn Foods · 2 yr 1 mo · Full-time · San Francisco, CA
- Built the trust & safety models now used in 7 product decisions a quarter.
- Taught SQL to 108 non-analysts.
- Cut false positives on payment blocks by 23%.
- Senior Data ScientistFeb 2021 - Sep 2024Cloudflare · 3 yr 7 mo · Full-time · San Francisco, CA
- Wrote the metric definitions the leadership team reviews every week.
- Taught SQL to 98 non-analysts.
- Senior Data ScientistNov 2018 - Dec 2020Apple · 2 yr 1 mo · Full-time · San Francisco, CA
- Replaced 63 hand-built dashboards with one metrics layer people trust.
- Cut false positives on payment blocks by 40%.
- Staff Data ScientistDec 2015 - Jul 2018Oracle · 2 yr 7 mo · Full-time · San Francisco, CA
- Caught $2 million of fraud in the first year of the new models.
- Senior Data ScientistJun 2012 - Dec 2015PepsiCo · 3 yr 6 mo · Full-time · San Francisco, CA
- Replaced 29 hand-built dashboards with one metrics layer people trust.
Projects
Fraud detection benchmark
2024Reproducible benchmarks comparing LightGBM and PyTorch on real workloads.
References
No references yet.
Education
- Northwestern UniversityBA, Economics2003 - 2007
Details
Compensation expectations
What they are targeting for their next role.
Visible to connections
Direct contact details
Direct email and a scheduling link.
Visible to connections