Skip to content
Segun UmoruSEGUN UMORU

Case Studies

Systems engineered for production.

Technical breakdowns of agentic AI architectures, real-time machine learning pipelines, and platform engineering projects built for enterprise scale and verifiable business impact.

Machine Learning·Optimus AI Labs
AI Engineering Lead (2025 – 2026)

OMNIS: 2D Financial Sentiment & Behavioral Fraud Intelligence Engine

Architected a dual-classifier financial intelligence engine scoring banking customers on Sentiment and Fit across a 20-zone strategic matrix, paired with a real-time Kafka event streaming fraud detection pipeline achieving 99%+ accuracy.

XGBoostApache KafkaPythonFastAPIScikit-learnMLflowRedisGreat Expectations

Verified Impact

  • 99%+ Fraud Detection Accuracy
  • 0.94-0.96 ROC AUC
  • Sub-10ms Inference Latency
  • 25% Churn Reduction
Read technical case study→
Platform Engineering·Blage Technologies Solutions
Senior AI/ML Engineer & Solutions Architect (2026)

Tage: Conversational Trading Assistant with Durable WhatsApp Ingress

Designed a resilient two-tier architecture pairing a NestJS core with a Python/FastAPI LangGraph AI engine, durable Twilio webhook queuing, multi-provider LLM failover, and persona-driven market analysis for African retail traders.

NestJSFastAPILangGraphBull/RedisPostgreSQLTwilioOpenAIAnthropic

Verified Impact

  • Zero dropped webhooks under load
  • 99.9% multi-provider LLM uptime
  • 5 distinct persona trading styles
Read technical case study→
Agentic AI·Blage Technologies Solutions
Senior AI/ML Engineer & Solutions Architect (2026)

TradeSense AI: Agentic Investment Assumption Monitoring Platform

Architected an autonomous multi-agent platform for VC and growth-equity funds that converts static investment memoranda into continuously verified, evidence-linked thesis graphs with durable execution and structural human-in-the-loop governance.

LangGraphTemporal.ioPythonFastAPINestJSPostgreSQLpgvectorDocker

Verified Impact

  • Sub-60s evidence brief generation
  • 100% architectural human-in-the-loop enforcement
  • Zero state loss on paused agent workflows
Read technical case study→
RAG & LLM·Vertex Software Corps
Senior AI/ML Ops Engineer (2026)

BUTLER: Enterprise Civic RAG Assistant with Automated PII Scrubbing

Architected and deployed a multi-language RAG chatbot for Dallas County government handling 5,000+ daily citizen inquiries with 99.9% availability, zero PII breaches across 10,000+ audit queries, and 40% latency reduction.

AWS BedrockOpenSearch ServerlessAmazon Titan 2Claude 3.5PythonFastAPI

Verified Impact

  • 5000+ Daily Queries
  • 99.9% API Availability
  • 100% PII Detection Accuracy
  • Zero 504 Timeout Errors
Read technical case study→
Machine Learning·Certin
Senior AI/ML Engineer (2026)

Predictive Logistics Anomaly Detection & SLA Breach Classification

Designed a multi-tenant machine learning platform combining per-carrier IsolationForest anomaly detectors with GradientBoosting SLA delay prediction classifiers, generating proactive breach alerts 24 to 48 hours in advance.

IsolationForestGradientBoostingLangGraphFastAPISQLAlchemy 2.0CeleryRedisDocker

Verified Impact

  • 10-20% Faster Issue Resolution
  • 10% Fewer Shipment Delays
  • 24-48h Advance SLA Warning
Read technical case study→

Have a complex technical challenge?

Let's evaluate your architecture, data readiness, and how to design agentic or predictive ML systems with real governance.

Start a conversation