Automated Deployment Across 29 Distribution Centers
McKesson's manual deployment process for their Drug Serialization Repository took four to six hours per location across 29 distribution centers, making DSCSA regulatory compliance nearly impossible to achieve on time. Improving automated the entire deployment pipeline using GitHub Actions, Kubernetes, Confluent Kafka, and MongoDB, reducing deployment time to 33 minutes per location and hitting the federal compliance deadline.
50%
Faster mean time to acknowledgement
5000+
Hours saved to automation
0
Missed incidents across 3 years
1 Billion Kafka Messages a Week with Zero Missed Incidents
Canada Post's internal IT team was overwhelmed managing a complex parcel tracking ecosystem across peak seasonal surges. Reactive troubleshooting, constant alerting, and mounting downtime threatened the reliability of the country's primary postal network. Improving built and managed a 24/7 production support framework on Azure, Kubernetes, and Kafka, handling over 3,300 incidents across three years with zero escalations to client stakeholders and zero missed incident acknowledgements.
Real-Time
Data routing and processing
Governed
Predefined schemas across every consumer
Self-Service
Teams route data without needing Kafka expertise
From Azure Event Hubs to Enterprise-Scale Kafka Streaming
Fastenal, the largest fastener distributor in North America, was struggling to route massive volumes of real-time data across systems using Azure Event Hubs — without a scalable architecture or internal Kafka expertise to build on. Improving designed and implemented an Event As A Service platform using Kafka Streams and Spring Boot that routes data in real time based on record metadata, scales without requiring Kafka knowledge from consuming teams, and was born directly from a solution presented at the Confluent Conference.
Agentic AI for Sales Enablement — From Manual Process to Production in 90 Days
Lakeshore Learning's sales team spent hours manually sourcing educational funding opportunities across government websites and spreadsheets. Improving built an Agentic AI system that crawls live data sources, identifies relevant opportunities autonomously, and hands off to sales with full context — deployed in 90 days.
Weeks to Days
Formulation challenge resolution time, down from weeks
B2B SaaS
MVP to full product launch, with paying customers
GPT-5
Powered by Azure OpenAI on Microsoft Azure Cloud
AI-Powered Co-Developer: From Scientific Journal to SaaS Product
The Institute of Food Technologists had decades of proprietary scientific knowledge locked in journal archives that food scientists couldn't practically access. Improving built IFT Co-Developer — an AI-powered SaaS platform featuring "Sous," an AI chatbot that gives food scientists instant access to IFT's knowledge base, solves complex formulation challenges, and generates customized ingredient recommendations. Taken from MVP to full B2B SaaS product launch.
White-Labeled
Scalable across Catalis's full client base
Serverless
No on-premise infrastructure required
Automated
Sensitive information redacted without manual review
AI-Powered Document Redaction — Manual Process to Cloud-Native Scale
Catalis was processing sensitive government documents on aging on-premise infrastructure that couldn't scale to meet client demand. Improving built a fully cloud-native redaction system on AWS — using Textract for OCR, Comprehend for text analysis, and serverless architecture that processes documents automatically and scales across departments without added overhead.
From COTS Dependency to Cloud-Native, Without Stopping Operations
Norwegian Cruise Line Holdings was locked into a rigid commercial booking system that took months to update and couldn't operate reliably at sea. Improving rebuilt the platform using Domain Driven Design and Reactive Architecture — a cloud-native, microservices system that ships features in days and keeps running when connectivity is gone.
2-3
Features deployed per sprint
5
Environment recovery time
5
Microservices deployed on Azure
From Oracle Licensing Lock-In to Cloud-Native
Berkshire Hathaway Energy inherited outdated Java Swing desktop applications and restrictive Oracle licensing after acquiring Dominion Energy's pipeline and processing facilities. Improving migrated the platform to a cloud-native React and Java microservices architecture on Azure — using Kafka as the messaging backbone and Terraform for infrastructure as code, so new environments spin up in minutes and failures don't take down the whole system.
50%
Reduction in modernization timeline
Offshore
Team enabled independently
Zero
Data exposure risk
Decades of Legacy ERP Knowledge Extracted in Weeks, Not Years.
Spring Point's decades-old ERP system had accumulated layers of undocumented business rules that only one developer fully understood — creating a modernization bottleneck that threatened to stretch the project beyond two years. Improving deployed Code Explorer, our proprietary AI-powered codebase analysis tool, giving their business analysts and offshore team a chatbot interface to query and understand legacy code without monopolizing the core developer's time.
Equipment issues surfaced before they cause downtime
Scalable
Expands across production lines and sites
Modern Real-Time Data Pipelines for Medical Device Manufacturing
Medtronic relied on manual entry and paper whiteboards to track production metrics across factory floors. Improving implemented Kafka and Confluent to capture live IoT and MES data in real time — enabling predictive maintenance, eliminating manual reporting delays, and giving production teams instant visibility into equipment health and line performance.
Centralized
Fragmented legacy systems unified into Databricks
HIPAA
& FedRAMP compliant governance built in
Self-Service
Engineers ingest, transform, and build independently
From Fragmented Legacy Systems to a Governed State-Wide Data Platform
DHCS was running behavioral health data across Excel, Microsoft Access, and a legacy Teradata warehouse — fragmented systems that made tracking health outcomes across California's Behavioral Health Transformation program nearly impossible. Improving built a centralized Databricks platform with Confluent pipelines, HIPAA and FedRAMP-compliant governance, and real-time BI reporting that gives DHCS data teams a self-service foundation they can actually build on.
Minutes
New environments spun up in minutes, down from weeks
Self-Service
Developers manage Kafka topics independently
3 Environments
Dev, test, and prod deployed on AWS with CI/CD
Confluent Kafka Platform — From Point-to-Point to Enterprise-Scale Streaming
Westfield needed scalable middleware for real-time data processing and future application development. Improving deployed Confluent Kafka across dev, test, and production environments on AWS and built a self-service API that empowered developer teams to manage Kafka topics and resources independently — without middleware dependency on every change.
