L3 Sustainment, SRE & Analytical Observability in Critical Platforms
Over a decade of my career was forged where production incidents carry immediate revenue impact, contractual SLA penalties, and operational disruption — spanning 5 years at BrScan Tecnologia and nearly 5 years at Serasa Experian. This case study synthesizes the operational architecture, Site Reliability Engineering (SRE) practices, Tier-3 Root Cause Analysis (L3 RCA) methodology, and custom Python (Pandas, Plotly, Dash) & PowerBI observability engines applied to sustain high-scale enterprise ecosystems.
Enterprise Scale Context
At Serasa Experian, I have provided technical leadership in sustaining and evolving mission-critical B2B platforms with dense synchronous and asynchronous integrations — including the GED360 suite (BrNotas, BrGestor, BrPronto, BrArquivo, SISGED), BrProntoPDV, and Portal TIM. Previously, at BrScan Tecnologia, I engineered operational intelligence pipelines for BrPronto PDV and BrFlow, and contributed directly to the Six Sigma Yellow Belt availability initiative for banking operations (BANRISUL and DJO 184).
L3 Incident Response, RCA & Change Governance Loop
Sustaining distributed enterprise systems under strict SLAs requires replacing ad-hoc firefighting with a closed-loop reliability and data-driven prevention pipeline:
Telemetry & Alerting
Logs, Monitors & Queues
SLA Shielding
Rapid L3 Mitigation
Deep RCA
Trace Correlation & Oracle DB
Fix Architecture
C4 Model, UML & BPMN
Change Governance
CAB / GMUD & Rollback
Preventive Intelligence
Python, Dash/Plotly & BI
Core Engineering Practices in Production
| Engineering Discipline | Recurring Enterprise Challenge | Technical Approach Applied |
|---|---|---|
Relational Diagnostics (Oracle DB) | Latency spikes during peak B2B transaction windows, lock contention, and heavy analytical reporting queries | Tuning complex SQL queries, diagnosing indexing and scan bottlenecks, and correlating DB sessions directly with application logs |
Integration Architecture (C4, UML, BPMN) | Fragmented tribal knowledge across satellite services and side-effect risks during evolutionary changes | Structured requirements engineering and living C4/UML/BPMN documentation mapping context boundaries, API contracts, and end-to-end flows |
Change Governance (CAB/GMUD) & SRE | Regression exposure during corrective and evolutionary rollouts within narrow maintenance windows | Pre-flight validation checklists, deterministic rollback runbooks, and automated resolution scripts for recurring Tier-3 incidents |
Sustainment Analytics Engines (Python, Pandas, Dash, PowerBI) | Limited visibility into chronic ticket recurrence, origin/status bottlenecks, and SLA drift by severity | Custom interactive web dashboards built in Python (Pandas, NumPy, Plotly, Dash) alongside PowerQuery/PowerBI executive Data Storytelling and Six Sigma Yellow Belt statistical analysis |
Turning Raw Incident Queues into Executive Engineering Action
Beyond deep database and code-level troubleshooting, I built automated analytical tools in Python (Pandas + Plotly + Dash) to transform raw operational support exports into actionable reliability intelligence:
- Chronic Offender & Recurrence Matrices: Algorithmic clustering of recurring incidents to prioritize structural code and database fixes targeting the 20% of root causes driving 80% of operational toil.
- Priority-Stratified SLA & Resolution Curves: Cross-correlating ticket origin, severity tier, and time-to-mitigate/resolve to guide executive sustainment reviews and continuous availability improvements.
- Public Clean-Room Engine (
Python+Dash+Plotly— 100% Synthetic Telemetry / Zero NDA): github.com/evslopes/sre-incident-analytics-dash
Inspect our clean-room public implementation featuring deterministic synthetic telemetry (Zero NDA), Pareto 80/20 recurrence fingerprinting, robust MAD Z-Score anomaly detection, reactive Dash console, and standalone HTML executive reports.