I turn messy data into decisions teams can act on.
Phoenix-based data analyst with 2+ years turning fragmented data into dashboards, SQL pipelines, and answers stakeholders actually use — with the Python, ML, and NLP chops to push past standard reporting when a problem calls for it.
role = data analyst
focus = SQL · BI dashboards · ETL · stakeholder reporting
degree = M.S. Information Technology, ASU · GPA 4.0
experience = 2+ yrs turning fragmented data into dashboards & decisions
edge = Python, ML & NLP projects that go beyond standard reporting
status = seeking data analyst & analytics-engineering roles
| # | domain | stack | focus |
|---|---|---|---|
| 0 | python | pandas · numpy · scikit-learn · pytorch · xgboost | modeling & analysis |
| 1 | cloud + mlops | sagemaker · lambda · emr · s3 · athena · cloudwatch · sns | deploy & automate |
| 2 | sql + data | sql · mongodb · snowflake · etl · stored procs | model & query |
| 3 | nlp + ai | biobert · regex ner · rag · transfer learning | extract meaning |
| 4 | viz + bi | tableau · power bi · matplotlib · seaborn | communicate |
| 5 | dev tools | javascript · angular · node · git · c++ | build the scaffolding |
Fine-tuned BioBERT on a Tesla T4 and bolted on a regex layer to catch what the model missed — pulling sugar-sweetened-beverage risk thresholds out of 167 PubMed abstracts. Recall climbed from basically zero to 77%, surfacing 28 high-confidence dose–response findings as clean, structured CSV.
A five-service AWS pipeline — S3 → EMR/PySpark → SageMaker → Lambda → SNS — that ingests wearable IoT data, scores cardiac risk with a tuned XGBoost model, and fires real-time alerts. Built in a schema-alignment guard so training and inference can't silently drift apart.
A fully normalized 3NF schema across seven tables that held zero integrity violations over 100+ records — with an audit trigger logging every change, three analytical views, stored procedures, a scalar UDF, and a cursor-driven glucose-alert routine.
Pulled and reshaped relational e-commerce data with heavy SQL — joins, CTEs, window functions — then built a Tableau dashboard on trends, regions, and product margin that pointed to a ~10% cross-sell opportunity.
Engineered 62 features across 6,187 records and deployed dual XGBoost models to a live SageMaker endpoint — then used CloudWatch to catch a false-positive bias and document the threshold-tuning fix instead of burying it.
Ten-plus seasons of IPL data explored with pandas, matplotlib, and seaborn to map player form, team strategy, and what actually swings a match.
I'm a data analyst who likes to get technical — the kind who learned early that a dashboard nobody trusts is worse than no dashboard at all. Most of my work is turning fragmented data into SQL pipelines, BI dashboards, and answers that a stakeholder can act on Monday morning.
What sets me apart from a typical analyst is how far I'll push when a problem calls for it: a BioBERT NLP pipeline pulling structured findings out of biomedical literature, a five-service AWS system scoring health risk in near real-time, a normalized SQL database that holds its integrity under load. I reach for Python, ML, and cloud tooling when standard reporting runs out of road — but the goal is always the decision, not the model.
I just finished my M.S. in Information Technology at Arizona State (4.0 GPA) and I'm looking for my next role — data analyst or analytics engineer — in Phoenix, remote, or anywhere in the US. If you've got messy data and a decision riding on it, I'd like to talk.