AI Products & Case Studies

Work We've Built & Shipped

End-to-end ML pipelines trained on exascale GPU infrastructure — real AI products, deployed and in production.

Case Study · Financial ServicesLive

Meridian Financial — AI Analytics Platform

End-to-end ML pipeline across 2.5M customer records, processed on exascale GPU compute. Four production models — churn prediction, credit risk, cross-sell propensity, and customer segmentation — surfacing $5.6B in actionable revenue opportunities.

Python · scikit-learnExascale GPU ClusterNext.js 16 · React 19Recharts · shadcn/uiVercel

Exascale Compute

NVIDIA A100/H100 · NVLink 4.0

InfiniBand HDR 400Gb/s

Low-latency node interconnect

2.5M Customer Records

Exascale data pipeline

4 Deployed ML Models

RF · GB · LR · K-Means

Interactive Dashboard

Customers Analyzed

0.0M

exascale pipeline

Total Assets

$0B

under management

Portfolio Churn Rate

0.4%

of customer base

Revenue Surfaced

$0.0B

cross-sell potential

68%

Model Accuracy

63.4%

AUC-ROC

32.4%

Portfolio Churn

14.4%

Top Feature Wt.

Churn Rate by Age Group

18–2530.7%
26–3530.9%
36–4531.6%
46–5532.9%
56–6534.2%
66+35.4%

Churn Rate by Satisfaction Score

Low (1–3)53.2%
Fair (3–5)42.6%
Good (5–7)35.2%
High (7–8.5)30.0%
Excellent (8.5–10)25.5%

Top predictor: account_age_months (14.4% importance)

Retention Recommendations

High-Complaint Customers

Proactive outreach with dedicated support rep

+70,000

est. saves

High

Single-Product Users

Cross-sell bundle with fee waiver incentive

+80,000

est. saves

High

Low-Engagement Digital

Onboarding program for mobile app features

+37,500

est. saves

Medium

Long-Dormant Accounts

Win-back campaign with competitive rate offers

+30,000

est. saves

Medium
Python · scikit-learn · pandas
Exascale GPU Cluster · A100/H100
Next.js 16 · Recharts · shadcn/ui
InfiniBand HDR · NVLink 4.0