Most data problems aren't data problems. They're process problems someone decided to solve with data.
I'm a Senior Data Engineer at ExponentHR working on the stack that HR and payroll systems run on. In the last year: compressed release cycles from 3 months to 14 days (the fix wasn't the code ; it was 11 weeks of cross-team idle time nobody was measuring), cut CDC pipeline compute costs by 67%, and built the Microsoft Fabric semantic layer that lets business teams ask data questions without filing a ticket. The systems I maintain hold a 98% uptime SLA. When they fail, people don't get paid.
Before this, at Missouri S&T, I built anomaly detection pipelines that filtered ~250 noisy P3 alerts per week down to what actually mattered and caught a production memory leak 4 hours before it took down a live system.
Outside work, I tend to build tools I wish already existed. Currently: an AI agent that monitors 109 company career pages around the clock, scores every new role against my resume, and pings me the moment something worth applying to goes live. It seemed more efficient than checking LinkedIn.
DP-700 Microsoft Certified Data Engineer. M.S. IST, Missouri S&T (4.0 GPA). Published researcher, Taylor & Francis 2025.
Specializing in CDC-based ETL pipelines, streaming data infrastructure, Microsoft Fabric semantic layers, and AI-powered data automation at production scale.
Data Engineer @ ExponentHR · Dallas, TX · July 2024 – Present
- Architected governed semantic layer on Microsoft Fabric with DAX metrics and AI agents, cutting support tickets ~40% and query response from 12s to under 4s
- Compressed deployment cycles from 3 months to 14 days via Azure DevOps CI/CD, restructuring cross-team handoffs that accounted for 11 weeks of idle time
- Reengineered CDC-based ETL from full-table reloads to incremental merge upserts, cutting batch runtime from ~30 min to under 8 min and compute costs by ~67%
- Owned payroll-critical recovery with containerized AAG failover, maintaining 98% uptime SLA with sub-hour restore targets
Data Engineer @ Missouri S&T · Rolla, MO · Aug. 2023 – July 2024
- Engineered Azure AI Anomaly Detector pipelines achieving 95%+ accuracy — caught a production memory leak 4 hours before outage
- Filtered ~250 weekly non-actionable P3 alerts via tunable thresholds, improving signal-to-noise from 1:5 to 1:1.2
- Migrated to AKS with HPA, consolidating 20 nodes to 4–8 dynamic, cutting monthly Azure spend by $3,200
- LinkedIn: linkedin.com/in/narendranathe
- Email:
- Portfolio: narendranathe.github.io
- GitHub: github.com/narendranathe
- Ask me about: Python, SQL, ETL/ELT, Azure Cloud, Spark, Power BI, Data Architecture, AI Automation
- All-time favorite film: The Dark Knight
- Published: Sentiment Analysis for Visitor Insights — Taylor & Francis, 2025 · DOI
- M.S. Information Science & Technology, Missouri S&T · GPA: 4.0/4.0
"Good data infrastructure is invisible — it runs, scales, and recovers without anyone noticing."