From raw data to a model you can trust.
Exploration, feature engineering, training, evaluation and practical model selection — always tied back to the business problem.
About / Cristina Varas
Data scientist building the bridge between models and real products.
I’m a Computer Engineer with a Master’s in Big Data & Analytics, working in Oracle’s OCTO EMEA at the intersection of Data Science, Machine Learning, AI and application delivery.
I spend most of my time turning complex technical possibilities into things people can actually test and use: predictive models, AI agents, vector search, intelligent applications, technical prototypes and customer-facing demos.
PROFILE / 01OCTO · EMEA
AI Workbench
Where data science becomes an AI system.
My workbench spans the whole path: explore the data, train and evaluate models, connect enterprise context, expose intelligence through APIs or applications, and make the result observable and understandable.
01ExploreSQL · notebooks · profiling
02Modelfeatures · ML · experimentation
03Evaluatemetrics · validation · explainability
04DeliverAPI · APEX · apps · demos
Agents need more than a prompt.
Memory, retrieval, vector search, enterprise data and orchestration turn a model into a useful AI workflow.
Intelligence should reach the user.
I connect models and AI services to APIs, Oracle APEX and other application experiences so the result can actually be consumed.
Build close to the data.
Oracle AI Database, OCI Data Science, Oracle Machine Learning and cloud services provide the platform layer behind the experiments.
How I work
Technical depth, without losing the story.
Start with the problem, not the model.
The right architecture comes after understanding the data, users, constraints and measure of success.
Prototype hands-on.
I like to get close to the actual data and technology: notebooks, SQL, Python, APIs, database capabilities and working application flows.
Design for production early.
Security, governance, deployment, observability and maintainability should not appear only after the demo works.
Explain what was built.
I turn complex implementations into diagrams, demos, articles and talks because useful technology has to be understandable.
Path
From software and analytics to applied AI.
My background combines Computer Engineering, Big Data & Analytics, predictive modelling, application development and cloud architecture. Before OCTO, I worked on analytics and predictive models and later on EMEA data-development projects, building automations, APEX experiences and technical demos.
Today that background lets me move comfortably between the model, the database, the application and the conversation with the people who need to use the result.
THE PRINCIPLE
Make AI useful and understandable.
For me, the interesting moment is not “we could build this”. It is when the data, model, architecture and user experience come together well enough for someone to say: this actually works.