2022
IT Support Analyst N1 (Intern)
First-line support for mission-critical operations across a national multi-brand food holding, focused on incident triage, resolution and retail continuity.
Tecnocomp
São Bernardo do Campo, BR
Data Engineer Building with AI · Python · PostgreSQL
I turn operations into information, and information into decisions.
I create solutions with data, automation and internal tools to improve processes, indicators and results.
I work at the connection between operational leadership, technology and business.
São Paulo, Brazil · UTC−3
- 32.000+
- tickets analyzed in BI
- −72%
- resolution time reduced with automation
- +37%
- team efficiency gain with automation
- 3 yrs
- from N1 intern to Lead Support Analyst
01 — Trajectory
From frontline support to technical lead work
2022
IT Support Analyst N2 — Infrastructure & Tooling
Technical escalation point for IT infrastructure across hundreds of business units nationwide, focused on standardization, network governance and internal tooling.
Tecnocomp
São Bernardo do Campo, BR
2024
Process Automation Engineer
Reduced operational bottlenecks through automation scripts and standardization of hybrid infrastructure routines.
Positivo Tecnologia
Curitiba, BR
2024
Senior Support Engineer
Led N1, N2 and Field Service support routines, focused on support quality, documentation, training, reliability and automation scripts.
Positivo Tecnologia
Curitiba, BR
2025 — now
Lead Support Analyst — Service Desk Operations, Data, BI & Automation
Operational leadership of the Service Desk team, connecting support routines with data pipelines, operational BI and automation.
Positivo S+
São Paulo, BR
CurrentNov 2025 — Present
Positivo S+
Lead Support Analyst — Service Desk Operations, Data, BI & Automation
- Lead Service Desk operations, aligning support routines, priorities, indicators and improvement actions with the needs of the business.
- Built an operational BI platform processing 21,000+ support records in production across enterprise contracts, fed by APIs, webhooks and CSVs with incremental ETL and SLA normalization.
- Built dashboards for volume, priority, status, owner, store, assignment group and critical category, adding period comparison, bottleneck detection and executive reading.
- Implemented a semantic incident-classification layer using local LLMs as support tooling, reducing miscategorizations and enriching unstructured support text.
- Delivered an enterprise knowledge platform with an integrated GPT-4 chatbot, 3-level RBAC, progressive brute-force protection and a 1,000+ event audit trail.
- Use AI-assisted development tools in daily workflows to accelerate diagnosis, analysis and implementation without positioning the work as AI model development.
PythonFastAPIDuckDBPolarsReact 19TanStack QueryNext.jsPostgreSQL
Nov 2024 — Nov 2025
Positivo Tecnologia
Senior Support Engineer
- Led N1, N2 and Field Service support routines, organizing documentation, training and technical guidance for the teams.
- Developed automation in PowerShell, Batch and Shell Script, delivering +37% team efficiency and −97% server-boot incidents where deployed.
- Structured the operational knowledge base and process documentation, shortening onboarding for N1, N2 and Field Service teams.
- Acted as escalation reference for complex incidents and ran technical training across support tiers.
- Implemented standardized procedures for hybrid Windows/Linux environments, improving compliance and operational stability.
PowerShellBatchShell ScriptVBScriptWindows ServerLinuxActive Directory
Jan 2024 — Nov 2024
Positivo Tecnologia
Process Automation Engineer
- Recovered 90% of the client's technology fleet through custom system imaging, reducing excessive hardware consumption and cutting provisioning time from 4h to 10–20 minutes per device, including servers.
- Engineered cross-platform scripts for critical task automation, removing manual toil and human error.
- Standardized hybrid Windows/Linux environments through governance scripts and configuration automation.
- Used prompt engineering and generative AI as support for technical diagnostics and code refinement in an AI-assisted workflow.
PythonPowerShellShell ScriptVBScriptGenerative AIPrompt Engineering
Nov 2022 — Jan 2024
Tecnocomp
IT Support Analyst N2 — Infrastructure & Tooling
- Built a modular Knowledge Base platform (HTML/CSS/JS) with indexed search and unique process codes, reducing recurring re-escalations.
- Cut average resolution time by 30% and raised team efficiency by 45% through automation and centralized documentation.
- Administered Windows Server environments and configured TCP/IP between POS terminals and central servers.
- Provided escalation support and remote technical guidance to Field Service teams.
Windows ServerLinuxTCP/IPHTMLCSSJavaScriptPowerShell
Jul 2022 — Nov 2022
Tecnocomp
IT Support Analyst N1 (Intern)
- Triaged and resolved incidents across ERP and POS systems for hundreds of business units.
- Bridged IT directives and field operations, keeping high-volume retail systems running.
- Supported organizational transitions during portfolio restructuring of the holding's brands.
ERPPOSWindowsTicketing Systems
02 — Selected Work
Systems, data pipelines and operational BI
Production work, internal tools and applied research with measurable operational impact.
01
HydraIA — water-risk intelligence for São Paulo
Academic ML · LLM EnsembleComputer Engineering capstone: a water-risk analysis system for the São Paulo Macrometropolis that fuses a time-series neural net with an ensemble of three LLMs under cross-validation.
- A bidirectional HydroLSTM forecasts reservoir levels from 26 years of official SABESP history (2000–2026).
- Three specialist LLMs (Gemini 2.5 Pro hydrology, GPT meteorology, local Mistral 7B synthesis) debate and cross-validate before a single report is emitted.
- Engineered with Strategy, Template Method and Dependency Injection — adding a new model is one class, zero orchestrator changes (Open/Closed).
- Hardened with atomic write-then-rename caching and covered by 58 unit tests.
- 26 yrs
- of training data
- 3 LLMs
- cross-validated ensemble
- 58
- unit tests
PythonPyTorchscikit-learnpandasGemini 2.5 ProGPTMistral 7BLM Studio
02
Operational BI for food-service support
Operational BI · SLA Analytics · AI-Assisted AnalysisA production intelligence platform for support operations, covering tickets, SLA and service performance across enterprise contracts.
- Consolidates data from APIs, webhooks and CSVs, normalizing operational fields into business indicators.
- Dashboards cover volume by period, heatmap by day and hour, SLA by priority, distribution by status, operation, assignment group, owner, store and critical category.
- Delivers executive reading of the period, comparison with the previous interval, bottleneck detection, most impacted stores and attention points for decision-making.
- Turns support operations into a management layer for prioritization, accountability and decision-making.
- 21,000+
- records analyzed
- API · CSV
- data ingestion
- SLA
- operational indicators
PythonFastAPIDuckDBPolarsPostgreSQLDockerReactLLMs
03
Enterprise knowledge platform with AI chatbot
Full-Stack · SecurityA secure enterprise knowledge-management platform with an integrated GPT-4 chatbot, 3-level RBAC and auditable internal usage.
- Integrated GPT-4 chatbot answering over governed internal knowledge.
- 3-level RBAC, progressive brute-force protection and a 1,000+ event audit trail.
- Built on Next.js and TypeScript with the OpenAI API.
- 3-level
- RBAC
- 1,000+
- audited events
Next.jsTypeScriptOpenAI APIPostgreSQL
04
RAG workflows & AI-assisted tooling
Retrieval · LLM Support · ToolingRetrieval and support tooling: semantic RAG plus guided LLM workflows connected to engineering and diagnostic routines.
- RAG engine with semantic chunking, pgvector storage and relevance evaluation.
- Guided workflows with LangChain and LangGraph for technical analysis.
- MCP Servers exposing local environments and tools safely to AI-assisted development tools.
- pgvector
- semantic retrieval
- MCP
- AI-assisted tooling
PythonLangChainLangGraphpgvectorMCP Servers
05
Low-level systems tooling
Go · Rust · CLIA line of focused utilities for Windows/Linux ecosystems, built to reduce repetitive operational work at the OS layer.
- go-file-processor — concurrent CSV-to-JSON converter built on Worker Pools for massive datasets.
- bar-minimal-tools — a Rust utility bar for Windows.
- A CLI toolkit in Go with concurrent networking and JSON formatting.
- Go · Rust
- concurrent by design
GoRustPowerShell 7
More on GitHub 03 — Expertise
Technical support, data and automation in the same workflow
From incident routines to management visibility.
01
Operational BI & Analytics Engineering
Dashboards and analytical views that turn support operations into management intelligence: SLA, volume, priority, status, owners, stores and critical categories.
Python · DuckDB · Polars · React
02
Data Pipelines
Pipelines from APIs, webhooks and CSVs to normalized datasets, consistent metrics and reliable analytical layers.
Python · FastAPI · DuckDB · Polars
03
AI-Assisted Data Workflows
LLMs used as support tools for classification, enrichment, interpretation and faster operational data reading.
LLMs · RAG · LangChain · MCP
04
Infrastructure & Automation
The foundation: years in 24/7 operations, automating repetitive work, standardizing environments and using data to reduce operational impact.
PowerShell · Docker · Power Automate
04 — Stack
The toolkit, grouped by intent
Languages
- Python95%
- TypeScript85%
- Go65%
- Rust55%
- JavaScript85%
- C / C++ / C#60%
- PowerShell 790%
Data & BI
- DuckDB90%
- Polars88%
- PostgreSQL85%
- APIs90%
- Webhooks85%
- CSV95%
- Operational BI92%
- SLA Analytics90%
LLM Support Tools
- LLMs (support)88%
- RAG82%
- pgvector75%
- LangChain80%
- LangGraph72%
- MCP Servers78%
- LM Studio80%
Frameworks
- Next.js85%
- React 1985%
- FastAPI88%
- Node.js82%
- TanStack Query78%
- Tailwind CSS88%
Infra & Corporate Automation
- Docker85%
- Linux85%
- Windows Server88%
- Power Automate85%
- Microsoft Teams80%
- InvGate85%
- Corporate APIs82%
AI-Assisted Development
- Claude Code92%
- Claude Designer80%
- ChatGPT 5.590%
- ChatGPT 5.5 Codex85%
- Cursor85%
- Gemini Pro80%
- Microsoft Copilot78%
IDEs
- VS Code95%
- Antigravity70%
- Cursor85%
05 — Certifications
A long, self-driven path
A selection from 35+ certificates — hardware benches to AI-assisted workflows.
Cloud & Strategy
- 2025
AWS Certified Cloud Practitioner
Amazon Web Services
- 2024
Leveraging Generative AI for Project Management
NASBA · PMI
Data & AI
- 2025
Imersão IA — 3ª Edição
Alura
- 2025
Imersão Dev — Agentes de IA
Alura · Google
- 2025
Imersão Dados com Python
Alura
- 2024
Imersão Dev com Google Gemini
Alura · Google
Backend & Data
- 2024
Node.js — REST APIs with Express & MongoDB
Alura
- 2024
ORM with Node.js — Sequelize & SQLite
Alura
- 2024
WebSockets — real-time with Socket.IO & MongoDB
Alura
- 2023
MongoDB — NoSQL, modeling & validation
Alura
Web & Front-end
- 2024
React — Design System with TailwindCSS
Alura
- 2023
HTTP — the web under the hood
Alura
- 2023
JavaScript for the Web — dynamic pages
Alura
Systems & Networks
- 2023
Computer Architecture — under your program
Alura
- 2023
Networks I–III — intranet to access policies
Alura
- 2023
Linux — terminal & shell
Alura
Where it started
- 2014
Computer Assembly & Repair
Technical course
- 2013
Networking & Web Design fundamentals
Technical course
06 — Contact
Let's turn operational routines into reliable data.
sousa3086@outlook.com