← All projects
PROFESSIONAL WORK / NOC OPERATIONS
From operational records to accountable decisions.
An auditable operational workflow built around deterministic preparation, bounded semantic review and transactional history, within the available Python and Microsoft 365 environment.
MY CONTRIBUTION
Workflow design, deterministic preprocessing, semantic-review contracts, UI execution, validation and evidence delivery.
TOOLS & METHODS
Python · regex · Copilot · Excel · SQL · Power BI · SharePoint · Power Automate
CHALLENGE
Operational records arrive as repeated ServiceNow snapshots, while tooling and access are constrained. The workflow needs to retain treatment history and produce outputs that can be inspected before they guide a decision.
DESIGN DECISIONS
Daily changes and occasional full-backlog snapshots remain separate evidence. Repeated incident records are retained as history. Deterministic calculations and field preparation run before bounded semantic review; missing or failed outputs are quarantined rather than accepted silently.
CONSTRAINTS
Repeated ServiceNow snapshots had to preserve history inside a constrained Python and Microsoft 365 environment. Deterministic rules ran before bounded semantic review; client records remain private.
The workflow, step by step.
01 / ServiceNow snapshots
Daily changes and occasional full backlog capture the operational record. Repeated incidents preserve treatment history.
02 / Path_Builder
Organizes period-based paths for inputs, outputs, quarantine and logs.
03 / build_agent
Extracts deterministic facts with regex; prepares calculations, bounded batches, tailored prompts and audit anchors.
04 / RPA + Compliance Agent
The UI executor checks coverage, retries, quarantines and resumes from anchors. Semantic review uses the indexed KB and the 34-column output contract.
05 / Consolidator
Checks successful days and anchors, then produces validated input, gaps, coaching and KPI views.
06 / Transactional load
Validates the column contract, sanitizes extra fields and normalizes values before an idempotent, all-or-nothing SQL load.
07 / SQL + export_backlog
RAW retains repeated records and source metadata. STG/MART derive active and historical views for reporting and sanitized exports.
08 / Intelligent Backlog
Consults incident dossiers, active backlog, history and daily Teams evidence. Sources and search limits stay visible; the analyst owns the decision.
IMPLEMENTATION & VALIDATION
Path_Builder separates input, output, quarantine and logs. build_agent prepares batches and audit anchors; UI automation checks coverage and resumes interrupted work. Consolidation validates gaps before idempotent, all-or-nothing SQL writes and RAW → STG/MART views.
RESULT & EVIDENCE
The supplied workflow produces validated Excel views, SQL history, executive Power BI data and sanitized inputs for Intelligent Backlog. Its declared scope is 68 KPIs across 12 operational domains; the semantic output contract contains 34 columns backed by an indexed 12-file knowledge base.
EVIDENCE LIMIT
This case describes the supplied professional workflow. It publishes no client records, internal endpoints or operational screenshots, and makes no claim of a live connection from this portfolio.
COMPLEMENTARY EVIDENCE
Power Automate normalizes daily Teams conversations as a side channel. Conversation evidence complements the incident record rather than replacing it.
RECOVERY AND CONTINUITY
Heartbeats, configurable retries, quarantine, audit anchors and a consecutive-failure limit support controlled execution, resumption and idempotency.
TRADE-OFFS & LESSONS
Regex and calculations improve repeatability but cannot replace contextual review. Semantic output needs schema checks, quarantined failures and analyst judgement. The declared 68 KPIs across 12 domains describe scope, not a throughput benchmark.
Explore architecture ↗
Next case: Incident correlation ↗
About the work and career ↗
VALTER LOURENÇO JUNIOR / NCMA SYSTEMS
← All projects
PROFESSIONAL WORK / NOC OPERATIONS
From operational records to accountable decisions.
An auditable operational workflow built around deterministic preparation, bounded semantic review and transactional history, within the available Python and Microsoft 365 environment.
MY CONTRIBUTION
Workflow design, deterministic preprocessing, semantic-review contracts, UI execution, validation and evidence delivery.
TOOLS & METHODS
Python · regex · Copilot · Excel · SQL · Power BI · SharePoint · Power Automate
CHALLENGE
Operational records arrive as repeated ServiceNow snapshots, while tooling and access are constrained. The workflow needs to retain treatment history and produce outputs that can be inspected before they guide a decision.
DESIGN DECISIONS
Daily changes and occasional full-backlog snapshots remain separate evidence. Repeated incident records are retained as history. Deterministic calculations and field preparation run before bounded semantic review; missing or failed outputs are quarantined rather than accepted silently.
CONSTRAINTS
Repeated ServiceNow snapshots had to preserve history inside a constrained Python and Microsoft 365 environment. Deterministic rules ran before bounded semantic review; client records remain private.
The workflow, step by step.
01 / ServiceNow snapshots
Daily changes and occasional full backlog capture the operational record. Repeated incidents preserve treatment history.
02 / Path_Builder
Organizes period-based paths for inputs, outputs, quarantine and logs.
03 / build_agent
Extracts deterministic facts with regex; prepares calculations, bounded batches, tailored prompts and audit anchors.
04 / RPA + Compliance Agent
The UI executor checks coverage, retries, quarantines and resumes from anchors. Semantic review uses the indexed KB and the 34-column output contract.
05 / Consolidator
Checks successful days and anchors, then produces validated input, gaps, coaching and KPI views.
06 / Transactional load
Validates the column contract, sanitizes extra fields and normalizes values before an idempotent, all-or-nothing SQL load.
07 / SQL + export_backlog
RAW retains repeated records and source metadata. STG/MART derive active and historical views for reporting and sanitized exports.
08 / Intelligent Backlog
Consults incident dossiers, active backlog, history and daily Teams evidence. Sources and search limits stay visible; the analyst owns the decision.
IMPLEMENTATION & VALIDATION
Path_Builder separates input, output, quarantine and logs. build_agent prepares batches and audit anchors; UI automation checks coverage and resumes interrupted work. Consolidation validates gaps before idempotent, all-or-nothing SQL writes and RAW → STG/MART views.
RESULT & EVIDENCE
The supplied workflow produces validated Excel views, SQL history, executive Power BI data and sanitized inputs for Intelligent Backlog. Its declared scope is 68 KPIs across 12 operational domains; the semantic output contract contains 34 columns backed by an indexed 12-file knowledge base.
EVIDENCE LIMIT
This case describes the supplied professional workflow. It publishes no client records, internal endpoints or operational screenshots, and makes no claim of a live connection from this portfolio.
COMPLEMENTARY EVIDENCE
Power Automate normalizes daily Teams conversations as a side channel. Conversation evidence complements the incident record rather than replacing it.
RECOVERY AND CONTINUITY
Heartbeats, configurable retries, quarantine, audit anchors and a consecutive-failure limit support controlled execution, resumption and idempotency.
TRADE-OFFS & LESSONS
Regex and calculations improve repeatability but cannot replace contextual review. Semantic output needs schema checks, quarantined failures and analyst judgement. The declared 68 KPIs across 12 domains describe scope, not a throughput benchmark.
Explore architecture ↗
Next case: Incident correlation ↗
About the work and career ↗
VALTER LOURENÇO JUNIOR / NCMA SYSTEMS