I lead Data, BI & AI at Spark Maxx, and there's only one bar here: systems in production. 15+ so far: dashboards, autonomous agents, analysis and local AI, each one documented on my portfolio with the problem, the decision, the architecture and what the choice cost.

Beyond that, I work as a fractional CAIO (Chief AI Officer) for selected companies: the executive function that owns AI strategy, governance and implementation, without the full-time seat. Open to B2B remote contracts (BR/US/Global).

---

How I work

Context before model. A powerful model without context is wasted potential, so I treat context as infrastructure: every project carries living documentation that AI reads before acting. It's how 15+ simultaneous systems fit one person.

Production is the bar. Every case is written as problem → decision → architecture → what the choice cost. A write-up is only done when the trade-off is stated.

🏗️ Data Engineering (GCP & MDS)	🧠 AI Architecture & Security	⚡ Cloud Infra & Edge MLOps
$\color{#4285F4}{\textsf{\textbf{GCP Data Stack:}}}$ BigQuery, Cloud Storage, Pub/Sub, Dataflow.	$\color{#9370DB}{\textsf{\textbf{AI-in-the-loop:}}}$ LLM-as-a-judge verification to block hallucinations and prompt injections.	$\color{#32CD32}{\textsf{\textbf{Serverless and IaC:}}}$ Immutable infrastructure as code and CI/CD pipelines.
$\color{#FF6C37}{\textsf{\textbf{Batch and ELT:}}}$ dbt (Data Build Tool) for dimensional modeling and Medallion Architecture.	$\color{#8A2BE2}{\textsf{\textbf{Agentic Workflows:}}}$ Autonomous workflows leveraging Claude / OpenAI APIs.	$\color{#E74C3C}{\textsf{\textbf{Deep Tech:}}}$ Over-The-Air (OTA) computer vision updates with zero-downtime inference.
$\color{#00CED1}{\textsf{\textbf{DAG Orchestration:}}}$ Idempotent, high-throughput message brokering (Airflow/n8n paradigms).	$\color{#DA70D6}{\textsf{\textbf{AI Privacy:}}}$ Data masking and PII sanitization for LLM integrations.	$\color{#1ABC9C}{\textsf{\textbf{Robotics:}}}$ Local telemetry caching and async Cloud sync for ROS2 nodes.

Has your data problem outgrown the spreadsheet? → victordataengineerds@gmail.com