Client retained at $14K/month · Phase 3: AI quality control on production floor (in progress)
Who we worked with
A US-based manufacturing company with 500+ employees operating multiple production facilities. They managed complex supply chain operations across several locations and processed thousands of purchase orders monthly, with annual revenue exceeding $80M.
Their 15-year-old ERP — built on .NET Framework 3.5 with a Windows Forms desktop client — had become a severe operational bottleneck blocking every modernization initiative the CTO wanted to pursue.
A 15-year-old system becoming a liability
The ERP had turned from a tool into a blocker — costing money, preventing growth, and failing security audits.
Crippling Performance
Report generation dragged on 12–15 minutes during peak periods. Decisions stalled, operations suffered.
12–15 min reportsDeployment Paralysis
Every update required 2–3 weeks of manual regression testing before it could ship. The system could not evolve.
2–3 wk deploy cycleManual Forecasting Bleeding Cash
3 full-time analysts ran inventory in Excel. Result: $400K+ excess inventory + $200K/quarter in lost sales from stockouts.
$600K+ annual impactZero Integration Capability
No API layer, no data pipeline — impossible to connect modern supply chain tools, e-commerce platforms, or ML models.
Unresolved Security Vulnerabilities
Flagged in two consecutive audits with no viable remediation path within the legacy architecture.
2 consecutive auditsAI Ambitions Blocked
The CTO wanted AI-powered demand forecasting — but the monolith had no API layer and no way to integrate ML models.
Two overlapping phases, zero downtime
A 22-week phased rollout — modernizing the core while building the AI layer simultaneously, with no single day of system disruption.
- Decomposed monolith into 14 microservices (.NET 8) — inventory, purchasing, production, shipping, reporting & more
- Migrated to AWS: RDS PostgreSQL, S3 (documents), ElastiCache Redis (sessions)
- Modern React frontend replacing Windows Forms — browser, tablet & mobile ready
- CI/CD via GitHub Actions — automated testing, staging & production release
- Containerized with Docker + AWS EKS (Kubernetes)
- Terraform IaC — full environment reproducible in under 30 minutes
- DataDog monitoring with custom real-time health dashboards
- Python + Apache Airflow data pipeline — historical sales, inventory, seasonal patterns & external signals (weather, regional events)
- Prophet demand forecasting model trained on 5 years of data — auto-generates purchase order recommendations
- Anomaly detection flags unusual inventory movements (theft, data errors, supplier issues)
- Executive dashboards: real-time AI insights — overstocked SKUs, demand spike predictions
- Forecasting fully embedded in purchasing module — PO recommendations automated end-to-end
Full-stack modern architecture
Results that changed how they operate
Measurable improvements across performance, cost, accuracy, and headcount — all delivered within the 22-week engagement.
Reports
from AI Forecasting
Reduction (Q1)
Accuracy
6 engineers, one cohesive delivery
A lean, high-output team structured for full-stack ownership — from cloud architecture to ML model training.
Solution Architect
System design, microservices decomposition, AWS architecture
Senior .NET Developer × 2
.NET 8 microservices, API layer, business logic migration
Senior React Developer
Modern frontend, executive dashboards, mobile-responsive UI
ML Engineer
Python, Prophet forecasting, Airflow pipelines, anomaly detection
DevOps / QA Engineer
CI/CD, Kubernetes, Terraform, DataDog, test automation
22-Week Full Delivery
Phased rollout — zero downtime, no single day of disruption during migration
What the client said
"The transformation i-verve delivered went far beyond what we expected from a modernization project. Our teams went from fighting the system every day to actually trusting it. The AI forecasting alone has changed how we think about procurement — we're no longer reacting to stockouts or sitting on excess inventory. This was a genuine operational transformation."