Data Scientist · Data Engineer · Analytics Engineer

Data Scientist · Data/Analytics Engineer

Building the pipeline. Building the model.

I bring 10+ years of SQL Server, SSIS, data warehousing, reporting, and production data operations, alongside recent professional work in BigQuery, dbt, and applied predictive analytics at UC Riverside. I connect data engineering and applied data science through data modeling, feature engineering, model evaluation, and stakeholder-facing reporting. M.S. Applied Data Science.

Oscar Gil Data
0Years Experience
M.S.Applied Data Science
0Conference Presentations
0Models in Early Alert pilot

What I bring

⌗

Languages & Query

PythonSQLR
☁

Cloud & Infrastructure

GCPBigQueryVertex AIDocker (AI-assisted use)dbt (UCR pilot)
◎

Machine Learning

Predictive ModelingSHAP (pilot analysis)scikit-learnPropensity-score matching (applied)
◈

Visualization

LookerSSRSMatplotlibPlotly
⊞

Data Engineering

SQL ServerSSRS / SSISETL PipelinesData Warehousing
≋

Analysis & EDA

pandasstatsmodelsSciPyNumPypytestDBeaver
∿

NLP & Text Classification

TF-IDF Vectorizationscikit-learn PipelinesText NormalizationDomain Abbreviation Mapping

Where I operate

Data Engineering
Current title
Data Science
Current title
Data Engineering
Data Science
AE Foundation

Data Engineering

Pipelines, modeling & reliability

dbt · 57-model pilot BigQuery ETL / ELT Star schema Docker (AI-assisted use) GCP Vertex AI Cloud Composer (AI-assisted DAG extension)
Strong overlap

Analytics Engineer

Metrics & transformation

dbt modeling (pilot) Dimensional modeling Feature engineering Looker SQL Data quality EDA Toolkit Kimball / SCD
Current title

Data Scientist

Modeling & inference

scikit-learn Random Forest SHAP (pilot analysis) McNemar's test Feature engineering Python Streamlit BQML
Professional experience
In progress ↗

Experience

Jan 2024 – Present University of California, Riverside

Data Scientist

Contributed to Early Alert predictive analytics by engineering behavioral features from LMS and SIS data, including engagement trends, activity rates, and temporal signals, and evaluating BigQuery ML Deep Neural Network (DNN) and Random Forest classifiers. Model evaluation included ROC-curve, recall, confusion-matrix, and threshold review. The project used DNN for academic-term weeks 1–4 and Random Forest for later weeks to support student-risk review; intervention impact was not measured.

Own source analysis, transformation requirements, SQL and data-modeling decisions, validation, troubleshooting, and operation for a 57-model Early Alert dbt pipeline in BigQuery. The pipeline transforms Banner and Canvas data into student-course-section-week feature models for stakeholder reporting. Its dbt implementation is substantially AI-assisted; the project is pilot-stage, designed for a weekly cadence, and manually triggered rather than automatically scheduled. A consultant initially set up the Early Alert Cloud Composer (Google Cloud managed Airflow) DAG environment; I later extended it with a second DAG phase and a reporting layer sequenced after prediction completes, using the established DAG-building and dependency approach.

Migrated consultant-built machine-learning applications into GCP-native notebook environments using Vertex AI, Docker, and Cloud Workstations. For Academic Personnel Office reporting, refactored legacy stored-procedure logic into warehouse-native SQL transformations, integrated Python-based statistical analysis, and developed a Looker dashboard for leadership reporting and operational transparency. The Academic Personnel Office nominated Oscar, and he later received the UCR STAR Award (2026).

Directed and validated an AI-assisted proof of concept for a retrieval-augmented generation (RAG) advising agent using Vertex AI Search & Conversation and a Gemini-powered Dialogflow CX Playbook Agent. Prepared Banner and Degree Works data in BigQuery for Vertex AI Search Data Stores and proposed a contextual helper-column pattern to make structured data more interpretable to the agent. The project included 34 AI-assisted parameterized dbt models and 114 automated dbt tests, all passing during validation; these tests support data/model quality checks, not agent-answer accuracy. Grouped approximately 14,220 student requirement rows into 1,159 distinct signatures and used templated calls instead of querying per student; the project record does not establish a measured API-cost comparison. This remains a pilot, with further development on hold pending internal review.

PythonBigQueryBigQuery MLVertex AIVertex AI SearchDialogflow CXCloud ComposerdbtDockerLookerSQLscikit-learnSHAP
2024 – Present Data Science Dynamics

Co-Founder

Co-founded Data Science Dynamics, a boutique data science practice focused on open-source tooling, applied ML research, and analytics engineering.

Co-developed the EDA Toolkit, an open-source Python library for exploratory data analysis and statistical workflows, under contract with Taylor & Francis (anticipated publication late 2026/early 2027). Presented EDA Toolkit at JupyterCon 2025 alongside collaborator Leon Shpaner (UCLA Health), demonstrating reproducible analytics pipeline design to an international audience.

PythonpandasSciPystatsmodelsSHAPscikit-learn
Apr 2026 – Present University of San Diego, Professional & Continuing Education

Lecturer I — Data Engineering

Facilitate asynchronous courses within USD PCE’s Practical Data Engineering certificate, including Practical Data Extraction and the Practical Data Engineering Capstone. Evaluate and grade student work, including end-to-end pipeline projects that use AWS services; respond to questions and provide technical feedback on data extraction, ingestion, data quality, reliability, scalability, and pipeline design. This is teaching and project evaluation, not professional AWS implementation experience. Guide students in applying production-oriented data engineering practices and communicating technical solutions to technical and non-technical audiences.

SQLPythonCloud Data PlatformsPipeline Design
2011 – 2023 National School District

Database Analyst

Served as Database Analyst throughout a 2011-2023 National School District tenure. Earlier in the role, contributed ETL code as a subject matter expert during the initial build of a consultant-architected, Kimball-style star schema data warehouse, working alongside the external consultant who led the architecture.

After the consultant relationship ended in 2017, took over full ownership of the warehouse, maintaining and extending SCD Type 1 and Type 2 change tracking across dimension and fact tables through 2023. Built and operated SSIS jobs through SQL Server Agent, including conditional failure-handling logic so non-critical step failures did not block the main nightly refresh, backed by automated failure alerting with diagnostic error output used to diagnose issues such as partner vendor SFTP key rotations or needed DDL changes.

Held direct SQL Server administration responsibilities, including user access, roles and permissions, performance tuning, and maintenance. Independently led two full SQL Server database migrations driven by hosting provider changes, with no data loss or extended downtime. Developed SSRS reporting, optimized SQL queries and BI infrastructure, and led stakeholder sessions to translate business requirements into structured analytical solutions and reporting frameworks.

SQL ServerSSISSQL Server AgentPythonSSRSETLStar SchemaData WarehousingDatabase Administration
Jan 2014 – Jul 2014 Fallbrook Union Elementary School District

Data Migration Consultant

Led a student information system migration project as an independent freelance consultant. Implemented a SQL Server environment to support the migration, automated SFTP ingestion for incoming source data, and generated migration-ready data files for the new student information system.

SQL ServerSFTPData Migration
Jan 2009 – Jun 2011 San Diego County Office of Education

Systems Technician II

Supported school districts with student information system data management, reporting, and systems integration work across varied district data environments.

SQLExcelData ManagementReportingSystems Integration

Portfolio

Data Engineering Case Study · Private Streamlit Application

Job Board Monitor

Developed a multi-source job intelligence and application workflow through AI-assisted implementation. The tool normalizes heterogeneous postings, classifies opportunities, evaluates fit, and maintains decision state across repeated scans. Oscar defined the workflow requirements, source coverage, data behavior, failure cases, and acceptance criteria, then reviewed and tested implementation changes, diagnosed integration failures, and operates the private application.

Data Pipeline Case Study · Applied Automation

Resume Screening Automation

Built a transparent Python pipeline to transform 388 unstructured PDF resumes into structured, reviewable candidate data under a four-day deadline. Extracted job-specific signals, applied explicit scoring rules, and produced a human-reviewable shortlist without claiming machine learning.

Portfolio Project · UCI Public Dataset

Student Dropout & Retention Predictor

This portfolio project uses the UCI student dropout dataset to demonstrate an analytics workflow: a flat file is transformed into a dimensional model with dbt and DuckDB, features are developed from exploratory analysis, and multiclass classifiers are evaluated in Python. The project reports a McNemar test result (p=0.0042), with Random Forest making fewer overall errors in that comparison, while Logistic Regression performs better for identifying at-risk enrolled students. These are project-reported results; their interpretation depends on the evaluation design and test assumptions. The project also presents model-performance analysis and SHAP-based prediction explanations in an interactive Streamlit demo. This is a portfolio application, not a production ML deployment.

Open Source Library · Under Contract Taylor & Francis

EDA Toolkit — PyPI Python Library

Co-developed an open-source Python library for fast, reproducible exploratory data analysis. Includes EDA tools, contingency table creation, hypothesis testing, and reproducible workflow helpers. Under contract with Taylor & Francis for publication (anticipated late 2026/early 2027).

Conference Presentation · 2025

JupyterCon 2025 — EDA Toolkit Tutorial

Co-presented a 25-minute session introducing EDA Toolkit to the JupyterCon audience — covering summary tables, automated profiling, distribution and crosstab visuals, and reproducible export workflows.

Graduate Team Capstone · 2022

Predicting ELPAC Proficiency for K–6 English Learners

Contributed to a graduate team capstone that modeled English proficiency (ELPAC) levels using five years of de-identified California school-district data. The project examined attendance and other features in relation to proficiency predictions and produced a Streamlit prototype for early identification; intervention impact was not evaluated.

Nonprofit / Pro Bono

Leadership Survey Data Analysis

Collaborated with a nonprofit via Catchafire to analyze pre- and post-retreat leadership survey data. Designed efficient data staging and visualization workflows to summarize participant feedback and support program evaluation.

Technical Guide

Detecting SQL Server Truncation Errors with Python

Troubleshot and resolved a SQL Server data truncation issue using SQL Agent monitoring and Python-based inspection.

Education & Credentials

Education

M.S. Applied Data Science

University of San Diego

B.S. Computer Networks

Coleman University

Certifications & Recognition

UCR STAR Award Recipient

Nominated by the Academic Personnel Office; recipient in 2026

2026

Taylor & Francis — EDA Toolkit (Under Contract)

Anticipated publication late 2026 / early 2027

Poster Presentation

“Reproducible Exploratory Data Analysis Frameworks for Pediatric and Lifespan Clinical Research,” presented jointly with Leon Shpaner at the Pediatric and Lifespan Data Science Conference

April 29, 2026 · Irvine, CA

JupyterCon 2025 Speaker

EDA Toolkit Tutorial on Streamlining Exploratory Data Analysis in Jupyter

November 4, 2025, San Diego, CA

Education Data Science Summit Presenter

Co-presented on ML & attendance modeling

2021

CTO Mentor Program Graduate (CCTO)

CITE, California IT in Education

Open to new
projects & opportunities

If you need help with data science, analytics, automation, SQL, Python, or report development, I'd love to hear about it.