Principal Engineer · Technical Leader

Madhan Selvam

Data · AI · Cloud · Architecture · Platform · Leadership

20+ years designing and modernizing enterprise data platforms — Lakehouses, data warehouses and hybrid cloud — for PepsiCo, State Street and Nike. I turn business problems into reliable, governed, AI-ready data products, and I lead the engineers who build them.

View my projects →Schedule a coffee chat

MSPlaceholder — add a headshot
(profile.headshot)

20+years building enterprise data, AI and cloud platforms

10+ PBof member data integrated from 30+ sources at Nike

350M+Nike members served by platforms I designed

$4M+cloud cost savings from platform optimization

80→99%reconciliation accuracy across PepsiCo systems

Professional Experience

Work and background

Two decades across retail, finance, healthcare and banking — from Mainframe ETL to cloud Lakehouses and AI-ready platforms.

Principal Engineer — Data & AI

Mar 2025 — Present

PepsiCo · Plano, TX

Principal architect setting technical direction across engineering, governance, product, BI and business stakeholders — translating business objectives into scalable technical solutions across enterprise marketing, integrated business planning and demand planning.

Azure DatabricksPySparkMedallion LakehouseSnowflakeUnity CatalogAI Agents

Principal Consultant — AML Sanctions

Jun 2024 — Mar 2025

State Street · Boston, MA

Led the modernization of an anti-money-laundering and sanctions platform from an on-prem Hadoop data lake to a governed AWS Databricks Lakehouse.

AWSDatabricksDelta LakeDelta Live TablesKafkaMWAA (Airflow)

Engineering Manager

Aug 2018 — Jun 2024

Nike · Beaverton, OR

Led data platform teams across two of Nike's largest analytics domains — global commerce and consumer (member) analytics.

AWSSnowflakeDatabricksSparkAirflowData Governance

Lead Data Engineer / Sr. Data Engineer — Consumer Analytics

Jan 2014 — Aug 2018

Nike · Beaverton, OR

Hands-on engineering of Nike's batch and streaming data platform during the move from Hadoop to Spark and the cloud.

HadoopSparkHivePrestoKafkaKinesis

Lead ETL Developer · Senior Consultant · Programmer Analyst · Software Engineer

May 2004 — Dec 2013

Aroghia · TCS · Cognizant · Sukraa · India/USA

A decade of ETL, data integration, data warehousing and Mainframe engineering for banking and healthcare.

InformaticaMainframeCOBOLJCLDB2Teradata

Read the full experience →

Insights

What twenty years of platforms taught me

The patterns that repeat across every role — each backed by outcomes, not adjectives.

Modernization without disruption

Every era of the stack, migrated live: Mainframe → Informatica → Hadoop → Spark → cloud Lakehouse. The constant is moving production platforms forward without stopping the business.

IBP integrations matured into enterprise capabilities across PBNA, PBUS and CAN at PepsiCo; CDH/HBase → AWS Databricks at State Street; Pig/Hive → PySpark at Nike.

Trust in the numbers

Platforms only matter if leaders believe the figures. Reconciliation, quality gates and a single source of truth come first.

80% → 99% reconciliation accuracy at PepsiCo; +25% reporting accuracy and −50% duplication at Nike Commerce.

Cost is an architecture decision

Performance tuning and storage design are treated as first-class engineering, not afterthoughts.

$4M+ savings at Nike; 70% faster and 40% cheaper storage at State Street; −30% batch runtime at PepsiCo.

Data that moves the business

Data products are measured by the decisions they change — forecasting, marketing spend, partner revenue.

$2B marketing spend optimized, $53B revenue visibility and $75M projected partner growth at Nike.

AI-ready platforms, AI-assisted teams

Building the governed foundations ML and GenAI need, and using AI to make engineering teams faster.

PepsiCo's "Sustain Agent" — MCP tools + RAG automating data-issue root cause analysis; A&M Hub Gold products built for AI-generated insights; 35% faster development with GitHub Copilot.