Case Studies

Data platforms, built for real stakes

These engagements reflect work delivered through Mudit's roles at Mphasis, Travelex, and orcablue.ai. The investment bank client is anonymized per standard confidentiality practice; Travelex and orcablue.ai are named as the platforms were built as internal analytics infrastructure for those organizations.

Building a Cloud-Native Data Product Factory for a Global Investment Bank

Client: Leading U.S. Investment Bank (anonymized) · Role: Delivery Module Lead — Data Product, via Mphasis

AWS Glue Athena Iceberg MWAA dbt Datadog

Context

The bank needed to scale analytics domain onboarding across Commercial & Investment Banking functions — without multiplying engineering effort for every new use case.

Challenge

Every new analytics domain required custom pipeline work. Business logic was duplicated and inconsistent across teams, with no standardized way to onboard new data products.

Approach

Architected a cloud-native Data Product Factory on AWS (Glue, Athena, Iceberg, MWAA). Designed reusable dimensional models and dbt transformation frameworks to standardize business logic, and led architecture, stakeholder alignment, and platform governance across banking functions.

1 Factory

Reusable framework replacing one-off pipeline builds

Std.

Standardized business logic across banking domains

Built-in

Observability via Datadog from day one

Consolidating Financial Analytics Across 250+ Global Retail Locations

Client: Travelex · Role: Senior Analyst, Data Engineering

Airflow CloudWatch Datadog Dimensional Modeling

Context

Travelex needed a single, trusted view of budget and sales performance across 250+ retail locations worldwide, previously assembled through manual, fragmented reporting.

Challenge

Manual consolidation across hundreds of locations was slow and error-prone. There was no consistent KPI definition across markets and limited visibility into pipeline health.

Approach

Architected a global financial analytics platform consolidating budget and sales data. Established observability, monitoring, and operational standards using Airflow, CloudWatch, and Datadog, and partnered with finance stakeholders to define KPI frameworks, metric governance, and data contracts.

80%

Reduction in manual reporting effort

250+

Retail locations consolidated onto one platform

Near real-time

KPI visibility across the business

Unifying E-Commerce & Marketplace Analytics Across Amazon, Flipkart, Nykaa & Unicommerce

Client: orcablue.ai · Role: Data Analyst

dbt Airflow Data Modeling Marketplace Integrations

Context

Enterprise e-commerce clients needed a single source of truth for sales and inventory across multiple marketplace platforms with different schemas and refresh cadences.

Challenge

Data was scattered across marketplace-specific systems with no standardized transformation layer, leading to slow, inconsistent reporting for enterprise clients.

Approach

Led development of scalable data ingestion and analytics frameworks for public, regulatory, and commercial datasets. Designed reusable transformation pipelines using dbt and Airflow to standardize data models across domains.

4 platforms

Amazon, Flipkart, Nykaa & Unicommerce unified

↑ Refresh

Improved data availability and refresh frequency

Reusable

Transformation pipelines standardized across domains

Reference Architecture Projects

Open, inspectable code — not just slides

Client work is confidential, so these are personal reference builds published on GitHub. They mirror the same architecture patterns — layered data platforms, governed data products, dbt-modeled transformations — applied to public datasets so anyone can read the actual code.

FX Data Platform — Enterprise Reference Architecture

A reference implementation of an enterprise-grade FX analytics platform, modeled on patterns used in real banking environments: layered ingestion → storage → transformation → governed data products, plus documented Architecture Decision Records (ADRs) and an observability & governance design.

AWS S3 Glue Redshift dbt Airflow Datadog ADRs

View on GitHub →

NYC

NYC Taxi Data Pipeline — AWS + Snowflake + dbt

An end-to-end ELT pipeline built on the Medallion architecture (Bronze → Silver → Gold), ingesting the public NYC Yellow Taxi dataset through S3 into Snowflake, with dbt-modeled transformation layers and a conceptual streaming variant using Kinesis Firehose and Glue.

AWS S3 Snowflake dbt Kinesis Glue Terraform

View on GitHub →

Want results like these for your platform?

Start with a Data Platform Health Check to see exactly where your architecture stands today.

Book a Health Check