Past Work

Real problems.
Measurable results.

A selection of data engineering and AI projects built across AWS — each one focused on a real business problem and a result you can point to.

Featured

GenAI Competitive Intelligence Agent

AI/ML

The Problem

Sales team spent hours manually researching competitors before each deal — no consistent, real-time data in CRM.

What Was Built

AI agent using AWS Bedrock + microservices that automatically delivers real-time competitive intelligence directly into Salesforce CRM.

50% reduction in research time. Improved seller win rates across AWS accounts.

60% efficiency improvement
AWS BedrockLambdaS3RedshiftPythonSalesforce
Featured

Cloud Data Lake Modernization

Data Engineering

The Problem

Legacy data infrastructure couldn't handle real-time event streams. Analytics ran on stale, batch-processed data.

What Was Built

End-to-end analytics solution with real-time streaming pipelines using EventBridge, Kinesis Firehose, and Glue — feeding a partitioned S3 data lake with Lake Formation governance.

Moved from batch to real-time. Full data governance layer. Scalable to any volume.

Legacy → modern data lake
AWS S3GlueAthenaKinesisLake FormationPython
Featured

Enterprise Analytics Dashboard Suite

Analytics

The Problem

Leadership had no unified view of business metrics. 30+ ad-hoc SQL queries run manually every week.

What Was Built

50+ complex SQL datasets and 30+ interactive dashboards with star schemas in AWS Redshift, surfaced via Amazon QuickSight.

40% faster query performance. Self-serve analytics — leadership gets answers without waiting on the data team.

40% faster query performance
AWS QuickSightRedshiftSQLTableau

Real-time Streaming Analytics Platform

Data Engineering

The Problem

Millions of events per day processed in nightly batches — meaning insights were always 24 hours stale.

What Was Built

Streaming analytics platform using Kinesis Data Streams, Lambda, and DynamoDB processing millions of events in real time.

Sub-second latency for real-time insights. Team moved from daily reports to live dashboards.

Sub-second event processing
KinesisLambdaDynamoDBCloudWatchPython

Automated Data Quality Framework

Data Engineering

The Problem

Downstream reports regularly failed due to bad source data — no automated validation layer in the pipeline.

What Was Built

Comprehensive data quality monitoring system using AWS Glue DataBrew and custom Python validators across enterprise datasets.

99.9% data accuracy. Downstream failures eliminated. Trust restored in the reporting layer.

99.9% data accuracy
AWS GlueDataBrewPythonCloudWatchSNS

AWS Cost Optimization Engine

Cloud Architecture

The Problem

Cloud costs growing faster than revenue. No visibility into which services or teams were driving spend.

What Was Built

Intelligent cost analysis tool using AWS Cost Explorer API, automated recommendations, and Tableau dashboards for spend visibility.

30% cost reduction. Engineering teams had clear ownership of their spend for the first time.

30% AWS cost reduction
AWS Cost Explorer APIPythonTableauLambda

Customer Sentiment Analysis Pipeline

AI/ML

The Problem

Customer feedback came in from 5+ channels — no unified view, no way to spot issues early.

What Was Built

End-to-end sentiment analysis pipeline using AWS Comprehend and custom ML models, processing feedback from all channels into a single dashboard.

Real-time sentiment visibility. Support team could surface and respond to issues hours earlier.

Real-time sentiment insights
AWS ComprehendSageMakerPythonElasticsearch

Enterprise Data Catalog

Data Engineering

The Problem

Engineers spent hours finding the right dataset — no central catalog, no metadata, no data ownership tracking.

What Was Built

Comprehensive data catalog using AWS Glue Data Catalog and custom metadata management, with a self-serve search UI.

70% improvement in data discovery time. Clear data ownership and lineage across the org.

70% faster data discovery
AWS GlueElasticsearchReactPythonGraphQL

Want results like these
for your data stack?

Let's talk about your current setup and what it would take to get your marketing data working as hard as you do.