Experience
Work & research
Production data platforms, applied ML, and the engineering that makes them useful.
VeYou
Data Engineer Intern
Delivered a governed BigQuery warehouse and reusable analytics for internal teams and partners.
Implementation details
- Designed and delivered a GCP-native ingestion and warehousing platform that consolidated numerous product, subscription, engagement, attribution, and backend data feeds into a governed BigQuery warehouse and data lake using Pub/Sub, Cloud Run, Cloud Storage, and BigQuery Data Transfer Service.
- Implemented a Medallion architecture in BigQuery with raw, standardized staging, and analytics-mart layers using version-controlled Dataform .sqlx models, CI/CD, idempotent loads, data-quality assertions, and schema-drift monitoring.
- Built internal and external partner-facing KPI dashboards in Looker Studio (Google Data Studio) and enabled governed NL2SQL access over curated lakehouse data, replacing manual cross-system reporting with reusable self-service analytics.
BigQueryDataformSQLXMedallionData LakeLooker StudioNL2SQLCloud RunPub/SubTerraformCI/CD
CUIMC
Data Engineer Intern
Cut PHI re-identification risk across 18+ entity types by building an NLP-based clinical de-identification pipeline.
Implementation details
- Enabled secure self-service access for five research teams by deploying governed datasets to AWS with role-based controls.
- Automated validation checks and documented end-to-end data lineage to keep outputs reproducible and auditable.
PythonClinical NLPDe-identificationAWSData ValidationData Lineage
Weatherhead East Asian Institute, Columbia University
Data Analyst Intern
Built automated Python ETL pipelines to clean, transform, and validate 2,200+ financial market records per day with 99%+ processing accuracy.
Implementation details
- Developed PostgreSQL data models and dbt-powered transformation workflows to support SQL analytics, trend analysis, benchmarking, and forecasting across 6+ months of historical data.
- Turned cleaned datasets into analyst-facing reporting and exploratory workflows for downstream decision-making.
PythonPostgreSQLdbtETLSQL AnalyticsForecasting
Cognizant Technology Solutions
Programmer Analyst
Reduced mean time to resolution (MTTR) by 40% across 250+ production cases by debugging GKE cluster failures through root-cause analysis.
Implementation details
- Improved reliability by building scripted anomaly detection and automated alerting that reduced monitoring overhead.
- Supported production workloads across GKE and Compute Engine, translating incident patterns into repeatable operational fixes.
GKEGCPPythonSREAnomaly DetectionRoot Cause Analysis