IT OPERATIONS · APPLIED AI · AIOPS
Valter Lourenço Junior.
Operational discipline. Applied to living systems.
12+ years in critical operations. Now building AI, automation and systems people can see, use and inspect.
Explore my work ↓
NCMA SYSTEMS / INDEPENDENT ENGINEERING LAB
EXPERIENCE BECOMES AN ENGINEERING METHOD
Built in operations. Applied to AI.
Understand the state. Reduce ambiguity. Control the change. Leave evidence behind.
2014–2022
Infrastructure & 24×7 operations
2022–2026
Major incidents & crisis
2026
Technical operations lead
2026 →
Applied AI systems
2022–2026 / SAME INCIDENT & CRISIS RESPONSIBILITIES / PROGRESSION FROM ANALYST TO SENIOR
SELECTED WORK / ACCENTURE / NOC OPERATIONS
Engineering under real constraints.
An audit-to-decision pipeline built with Python, Copilot, Excel, SQL and native Microsoft 365 connectors under restricted tool access.
68
KPIS
12
DOMAINS
01
Prepare context
ServiceNow snapshots / Path_Builder / build_agent
02
Execute & audit
RPA / Compliance Agent / Consolidator
03
Commit & derive
Transactional SQL RAW / STG / MART
04
Deliver evidence
Power BI / SharePoint / Intelligent Backlog
MY CONTRIBUTION / Deterministic preparation, semantic review, validation and evidence delivery.
Inspect the case ↗
Read the full case ↗
PROFESSIONAL WORK / INCIDENT CORRELATION
Less investigation. More time to operate.
The audited backlog becomes an evidence layer for contextualization, correlation and analyst support. Sources and search limits stay explicit.
~90%
LESS CONTEXTUALIZATION TIME
5–15 min
MANUAL CONTEXT
<30 sec
AUTOMATED CONTEXT
≈4 h / shift
TIME RELEASED
Approximate operational impact reported in my professional work.
INTELLIGENT BACKLOG / EVIDENCE CASCADE
C1
Incident dossiers
Active incident .md files / SharePoint
C2
Active backlog
Active backlog CSV + export anchor
C3
History + Teams
Historical CSV + daily conversation CSV
Retrieval remains probabilistic. Missing evidence triggers an explicit limit and ServiceNow verification. Sources and the search path are disclosed; the analyst decides.
Inspect the case ↗
Read the full case ↗
PROFESSIONAL PRACTICE / MAJOR INCIDENT & CRISIS
Incident Commander. Clarity under pressure.
I led P1/P0 war rooms, assessed the initial context and mobilized technical teams across infrastructure, applications, networks, DBA and suppliers.
Alongside incident leadership, I owned resource coordination, timelines and criticality-based updates for stakeholders, leadership and clients.
ONE ANALYST / SIX INCIDENT FUNCTIONS
01
Incident Lead
02
Communications Lead
03
Scribe / timeline
04
Resource coordination
05
Stakeholder liaison
06
Escalation / process
TIMELINES / RCA / EXECUTIVE COMMS / POWER AUTOMATE / COPILOT / GENAI
Read the full case ↗
NCMA SYSTEMS / INDEPENDENT ENGINEERING LAB
The lab extends the discipline.
Local and cloud models, grounded knowledge, tools and deterministic gates. Independent systems built around observable state and human control.
01 / THE LAB, IN MOTION
From intent. To systems you can see.
Real interfaces. Connected infrastructure. A closer look at what I build.
NCMA SYSTEMS / A SHORT STUDY Interfaces, captures and code from my work. Jarvis shown in sample mode.
JARVIS / PERSONALAGENT / VOICE, MEMORY AND TOOLS
A voice. A presence. A system behind it.
Idle / original Jarvis UI
Silent visual demonstration · runtime not connected
One intent. Multiple runtimes.
A local-first personal intelligence system connecting voice, memory, knowledge, MCP tools and web research.
The orb follows speech amplitude, with distinct idle, listening and speaking states. A visible response to voice keeps the interaction legible.
Local and cloud inference are resource-aware choices. Presentation state stays separate from permission to execute tools.
Inspect Jarvis ↗
See the runtime boundaries ↗
INPUT
Voice / text
POLICY
Local / cloud routing
CONTEXT
Memory / MCP / research
AUTHORITY
Guarded execution
Read the full case ↗
LOCAL AGENT LAB / LOCAL LLM ORCHESTRATION
Local reasoning.
Verified progress.
Model outputs are proposals. Independent source validation turns them into progress a local agent can trust, checkpoint and resume.
PUBLISHED G4H RESULTS
One synthetic deterministic fixture
- Cases correct
- 12/12
- Fields correct
- 60/60
- Checkpoint rounds
- 22
G4H / FROM EVIDENCE TO ACCEPTED STATE
- 01
Retrieve evidence
Focused authoritative source rows.
- 02
Propose a next step
HIGH draft. MEDIUM JSON finalizer.
- 03
Validate independently
Reject unsupported graph transitions.
- 04
Checkpoint and resume
Keep verified progress across rounds.
KNOWLEDGE RUNTIME / ORACLE AI DATABASE 26AI
Answers with a way back.
Hybrid retrieval, source attribution and a controlled knowledge lifecycle. Sources remain available as information becomes an answer.
01 / SOURCE
Collect evidence.
02 / CHUNK
Keep usable context.
03 / EMBED
Represent meaning.
04 / RETRIEVE
Find relevant evidence.
05 / ANSWER
Retain the sources.
06 / VALIDATE
Apply deterministic gates.
Illustrative provenance / complete sequence remains legible without motion.
Explore the knowledge path ↓
Read the full case ↗
MODFORGE / AI-ASSISTED ENGINEERING
Generation is a candidate. Evidence makes a release.
Contracts, bounded generation, deterministic validation and build evidence. Human Plane governs effective state, review and guarded mutations.
01
Specification
02
Architecture
03
Generation
04
Validation
05
Build
06
Evidence
07
Settlement
HUMAN REVIEW / EXPLICIT AUTHORIZATION
Effective state · guarded mutations · evidence · blast-radius awareness
Cyberpunk is an applied engineering domain. Explore the existing collections at /forge ↗.
WEB RESEARCH / SEARXNG + CRAWL4AI
Research with sources attached.
Self-hosted search and extraction on private infrastructure. Source collection, synthesis and citations form one inspectable research path.
Search
Extract
Sources
Synthesis + citations
Infrastructure and extraction validated / physical synthesis smoke remains pending.
Read the full case ↗
03 / OPERATIONAL VISIBILITY
A view over recorded history.
A workflow snapshot shows how the Control Room reads dashboard history, renders the interface and responds to a request.
01
GET Control Room
02
Read Dashboard History
03
Render Control Room
04
Respond Control Room
ENTRY
The first node is a webhook entry for the Control Room view.
READ PATH
The next node reads dashboard history from a data table. A code node renders the view, followed by a response node.
EVIDENCE LIMIT
The snapshot establishes node types and their connections. It contains no live telemetry, payloads, credentials or proof of current service availability.
INSPECTION OVER METRICS
This portfolio presents the recorded workflow topology. It does not invent uptime, incident counts or performance results.
Topology snapshot dated 27 September 2026. Only node names, types and connections are used; operational parameters are excluded.
Read the full case ↗
VALTER LOURENÇO JUNIOR
A career in systems. A bias for evidence.
IT operations, service management, major incidents and crisis work shape how I build AI: observable state, bounded actions and evidence people can review.
Applied AI postgraduate studies ESIC × Coders · In progress
Systems Analysis and Development Estácio · Completed 2023
ITIL 4 Foundation · PeopleCert
Explore all projects ↗
Explore all architectures ↗
IT OPERATIONS · APPLIED AI · AIOPS
Valter Lourenço Junior.
Operational discipline. Applied to living systems.
12+ years in critical operations. Now building AI, automation and systems people can see, use and inspect.
Explore my work ↓
NCMA SYSTEMS / INDEPENDENT ENGINEERING LAB
EXPERIENCE BECOMES AN ENGINEERING METHOD
Built in operations. Applied to AI.
Understand the state. Reduce ambiguity. Control the change. Leave evidence behind.
2014–2022
Infrastructure & 24×7 operations
2022–2026
Major incidents & crisis
2026
Technical operations lead
2026 →
Applied AI systems
2022–2026 / SAME INCIDENT & CRISIS RESPONSIBILITIES / PROGRESSION FROM ANALYST TO SENIOR
SELECTED WORK / ACCENTURE / NOC OPERATIONS
Engineering under real constraints.
An audit-to-decision pipeline built with Python, Copilot, Excel, SQL and native Microsoft 365 connectors under restricted tool access.
68
KPIS
12
DOMAINS
01
Prepare context
ServiceNow snapshots / Path_Builder / build_agent
02
Execute & audit
RPA / Compliance Agent / Consolidator
03
Commit & derive
Transactional SQL RAW / STG / MART
04
Deliver evidence
Power BI / SharePoint / Intelligent Backlog
MY CONTRIBUTION / Deterministic preparation, semantic review, validation and evidence delivery.
Inspect the case ↗
Read the full case ↗
PROFESSIONAL WORK / INCIDENT CORRELATION
Less investigation. More time to operate.
The audited backlog becomes an evidence layer for contextualization, correlation and analyst support. Sources and search limits stay explicit.
~90%
LESS CONTEXTUALIZATION TIME
5–15 min
MANUAL CONTEXT
<30 sec
AUTOMATED CONTEXT
≈4 h / shift
TIME RELEASED
Approximate operational impact reported in my professional work.
INTELLIGENT BACKLOG / EVIDENCE CASCADE
C1
Incident dossiers
Active incident .md files / SharePoint
C2
Active backlog
Active backlog CSV + export anchor
C3
History + Teams
Historical CSV + daily conversation CSV
Retrieval remains probabilistic. Missing evidence triggers an explicit limit and ServiceNow verification. Sources and the search path are disclosed; the analyst decides.
Inspect the case ↗
Read the full case ↗
PROFESSIONAL PRACTICE / MAJOR INCIDENT & CRISIS
Incident Commander. Clarity under pressure.
I led P1/P0 war rooms, assessed the initial context and mobilized technical teams across infrastructure, applications, networks, DBA and suppliers.
Alongside incident leadership, I owned resource coordination, timelines and criticality-based updates for stakeholders, leadership and clients.
ONE ANALYST / SIX INCIDENT FUNCTIONS
01
Incident Lead
02
Communications Lead
03
Scribe / timeline
04
Resource coordination
05
Stakeholder liaison
06
Escalation / process
TIMELINES / RCA / EXECUTIVE COMMS / POWER AUTOMATE / COPILOT / GENAI
Read the full case ↗
NCMA SYSTEMS / INDEPENDENT ENGINEERING LAB
The lab extends the discipline.
Local and cloud models, grounded knowledge, tools and deterministic gates. Independent systems built around observable state and human control.
01 / THE LAB, IN MOTION
From intent. To systems you can see.
Real interfaces. Connected infrastructure. A closer look at what I build.
NCMA SYSTEMS / A SHORT STUDY Interfaces, captures and code from my work. Jarvis shown in sample mode.
JARVIS / PERSONALAGENT / VOICE, MEMORY AND TOOLS
A voice. A presence. A system behind it.
Idle / original Jarvis UI
Silent visual demonstration · runtime not connected
One intent. Multiple runtimes.
A local-first personal intelligence system connecting voice, memory, knowledge, MCP tools and web research.
The orb follows speech amplitude, with distinct idle, listening and speaking states. A visible response to voice keeps the interaction legible.
Local and cloud inference are resource-aware choices. Presentation state stays separate from permission to execute tools.
Inspect Jarvis ↗
See the runtime boundaries ↗
INPUT
Voice / text
POLICY
Local / cloud routing
CONTEXT
Memory / MCP / research
AUTHORITY
Guarded execution
Read the full case ↗
LOCAL AGENT LAB / LOCAL LLM ORCHESTRATION
Local reasoning.
Verified progress.
Model outputs are proposals. Independent source validation turns them into progress a local agent can trust, checkpoint and resume.
PUBLISHED G4H RESULTS
One synthetic deterministic fixture
- Cases correct
- 12/12
- Fields correct
- 60/60
- Checkpoint rounds
- 22
G4H / FROM EVIDENCE TO ACCEPTED STATE
- 01
Retrieve evidence
Focused authoritative source rows.
- 02
Propose a next step
HIGH draft. MEDIUM JSON finalizer.
- 03
Validate independently
Reject unsupported graph transitions.
- 04
Checkpoint and resume
Keep verified progress across rounds.
KNOWLEDGE RUNTIME / ORACLE AI DATABASE 26AI
Answers with a way back.
Hybrid retrieval, source attribution and a controlled knowledge lifecycle. Sources remain available as information becomes an answer.
01 / SOURCE
Collect evidence.
02 / CHUNK
Keep usable context.
03 / EMBED
Represent meaning.
04 / RETRIEVE
Find relevant evidence.
05 / ANSWER
Retain the sources.
06 / VALIDATE
Apply deterministic gates.
Illustrative provenance / complete sequence remains legible without motion.
Explore the knowledge path ↓
Read the full case ↗
MODFORGE / AI-ASSISTED ENGINEERING
Generation is a candidate. Evidence makes a release.
Contracts, bounded generation, deterministic validation and build evidence. Human Plane governs effective state, review and guarded mutations.
01
Specification
02
Architecture
03
Generation
04
Validation
05
Build
06
Evidence
07
Settlement
HUMAN REVIEW / EXPLICIT AUTHORIZATION
Effective state · guarded mutations · evidence · blast-radius awareness
Cyberpunk is an applied engineering domain. Explore the existing collections at /forge ↗.
WEB RESEARCH / SEARXNG + CRAWL4AI
Research with sources attached.
Self-hosted search and extraction on private infrastructure. Source collection, synthesis and citations form one inspectable research path.
Search
Extract
Sources
Synthesis + citations
Infrastructure and extraction validated / physical synthesis smoke remains pending.
Read the full case ↗
03 / OPERATIONAL VISIBILITY
A view over recorded history.
A workflow snapshot shows how the Control Room reads dashboard history, renders the interface and responds to a request.
01
GET Control Room
02
Read Dashboard History
03
Render Control Room
04
Respond Control Room
ENTRY
The first node is a webhook entry for the Control Room view.
READ PATH
The next node reads dashboard history from a data table. A code node renders the view, followed by a response node.
EVIDENCE LIMIT
The snapshot establishes node types and their connections. It contains no live telemetry, payloads, credentials or proof of current service availability.
INSPECTION OVER METRICS
This portfolio presents the recorded workflow topology. It does not invent uptime, incident counts or performance results.
Topology snapshot dated 27 September 2026. Only node names, types and connections are used; operational parameters are excluded.
Read the full case ↗
VALTER LOURENÇO JUNIOR
A career in systems. A bias for evidence.
IT operations, service management, major incidents and crisis work shape how I build AI: observable state, bounded actions and evidence people can review.
Applied AI postgraduate studies ESIC × Coders · In progress
Systems Analysis and Development Estácio · Completed 2023
ITIL 4 Foundation · PeopleCert
Explore all projects ↗
Explore all architectures ↗