Ganesh Makkina

About Ganesh Makkina

I build data systems that turn complex, high-volume financial workflows into reliable internal products teams can use independently.At S&P Global, I work on a pipeline that ingests, extracts, transforms, and standardizes thousands of SEC filings (10-K and 10-Q) for downstream analyst consumption. The system processes XBRL JSON filings, extracting millions of financial facts across hundreds of companies into Databricks Delta tables — running as scheduled production jobs with multi-threaded orchestration and automatic retry logic.The original pipeline was built on PySpark and Spark SQL in Databricks notebooks — heavy on cluster compute and tightly coupled to the Spark runtime. I led the effort to rewrite the extraction and transformation engine in pure Python, replacing Spark DataFrame operations with dictionary-based processing. The result: 90%+ reduction in per-file processing time, no Spark cluster dependency, and a pipeline that runs anywhere — Databricks, Docker, or Kubernetes — with zero code changes.I built a multi-stage validation framework that compares output across pipeline versions at every layer — master records, metadata, and financial facts — with normalization rules that separate real discrepancies from representation-only differences. The rewrite was validated row-by-row across tens of thousands of facts with exact parity confirmed.The transformation layer maps raw XBRL tags to standardized S&P mnemonics using a configurable rules engine with conditional logic, multi-column matching, and expression evaluation.On the productization side, I drove end-to-end delivery:1. Containerized with Docker (Flask API + FastAPI Sandbox UI)2. Deployed to Kubernetes on AWS EKS via Azure DevOps CI/CD with Helm charts3. Integrated HashiCorp Vault for zero-trust credential management4. Built a Sandbox UI for analysts to browse tables, compare mapping rules, trigger runs, and promote validated changes — no engineering support needed5. Designed for environment portability: all config via env vars, one Helm values file per environmentThe system now serves as the foundation for corporate XBRL financial data at S&P, with clean separation between extraction, transformation, and delivery.Core strengths: Data Engineering · Python · Databricks · SQL · ETL/ELT · Docker · Kubernetes · Flask · FastAPI · CI/CD · AWS EKS · Vault · System Design

Experience

1. Senior Data Engineer

   S&P Global

   Oct 2025 — Present · NY, US

   Building a financial data pipeline that extracts, transforms, and standardizes SEC filings (10-K/10-Q) at scale, processing XBRL JSON filings and extracting millions of financial facts into structured Databricks Delta tables for downstream analyst consumption- Rewrote the extraction engine from PySpark/Spark SQL notebooks to a pure Python pipeline using dictionary-based processing — achieving 90%+ reduction in per-file processing time and eliminating the Spark cluster dependency entirely- Built a configurable transformation layer that maps raw XBRL tags to standardized S&P financial mnemonics using a rules engine with conditional logic, multi-column matching, and expression evaluation- Designed and implemented a multi-stage validation framework comparing pipeline output across versions at the master, metadata, and financial fact level — confirming row-by-row parity across tens of thousands of records- Deployed the full system to Kubernetes on AWS EKS via Azure DevOps CI/CD with Helm chart-driven configuration and HashiCorp Vault for secure credential injection- Built an internal Sandbox UI (FastAPI) that enables analysts to browse output tables, compare baseline vs. modified mapping rules, trigger pipeline runs on demand, and promote validated changes — removing the need for engineering involvement in day-to-day operations- Containerized the application with Docker (Flask health API + FastAPI UI), designed for environment portability with all configuration driven via environment variables and a single Helm values file per environment.

Education

• Sri Gayatri Educational Institutions

  10+2, MPC

• Army Institute of Technology, Pune

  Bachelor of Engineering, Computer Science

  2017 — 2021

• Rutgers University–New Brunswick

  Master of Science - MS, Statistics - Data Science

• Defence Laboratories School

  10th, CBSE