Shivam Pareek

About Shivam Pareek

I build data infrastructure that moves fast and breaks nothing.Over 5 years, I\'ve designed and shipped production-grade pipelines, lakehouses, and MLOps systems at scale — currently processing 3TB+ of daily financial data at KPMG India on Azure Databricks and Delta Lake.What I actually do:→ Design Medallion Architecture lakehouses (Bronze → Silver → Gold) from scratch, not templates→ Build PySpark pipelines optimized at the job level — partition tuning, broadcast joins, Z-ordering — not just \"working\" but fast→ Ship ML models to production with MLflow: cut deployment time 40%, reduced compute costs 30%→ Own the full stack: data ingestion → transformation → serving → monitoring → CI/CDRecent work I\'m proud of:• RAG-based Document Intelligence Pipeline using LangChain, Azure OpenAI, and pgvector — built end-to-end, handles enterprise-scale document retrieval• Real-Time Lakehouse on Azure Databricks with Unity Catalog governance across 20+ datasets, column-level security, full lineage• Retail forecasting platform at Tiger Analytics: 5M+ SKUs, 35% faster Spark jobs, 50% query improvement on Delta LakeStack: Python · PySpark · SQL · Azure Databricks · Delta Lake · ADLS Gen2 · ADF · MLflow · Unity Catalog · LangChain · pgvector · AWS (S3, Glue, Redshift, EMR) · Spark Streaming · CI/CD · GitI\'m selective about my next move — looking for senior data engineering or staff-level roles where the data problems are hard and the engineering bar is high.Open to: Google · Meta · Netflix · Amazon · or any team doing serious distributed systems work at scale. s••••••••@gmail.com

Experience

1. Data Engineer

   KPMG India

   Jul 2025 — Present · Gurugram, IN

   Building production-grade data infrastructure at KPMG India — 3TB+ of daily financial data processed across a Medallion Architecture lakehouse on Azure Databricks.What I own:→ End-to-end MLOps platform: MLflow experiment tracking, model registry, automated deployment — cut time-to-production from weeks to days (40% faster)→ Unity Catalog governance: column-level security, data lineage tracking across 20+ datasets→ PySpark ETL/ELT pipelines for regulatory reporting and analytics — built for correctness, not just throughput→ Databricks cluster optimization: auto-scaling + spot instances → ~30% reduction in monthly compute spend→ CI/CD and DataOps: Git branching strategy + automated pipeline testing → 60% fewer deployment failures→ Cross-functional ownership: partnered with data science to productionise

Education

• Sobhasaria Engineering College

  Bachelor of Technology - BTech, Mechanical Engineering

  2016 — 2020

• PRINCE ACEDEMY OF HIGHER EDUCATION

  XII, Science