Headline

Data Analyst | Turning raw data into decisions - 0.87 ROC-AUC risk model, 40% faster SQL pipelines | SQL · Python · Power BI · AWS | ETL & ML

About

Most dashboards get built, get admired in one meeting, and then never opened again. Most "risk models" live in a notebook and never see production. That's the gap I kept running into - analysts who can describe data, but can't get anyone to actually act on it. That changed for me when I built a credit risk model that wasn't just accurate on paper - it had to explain why someone was flagged high-risk, in a way a non-technical stakeholder could trust. That's when I stopped thinking like someone running queries, and started thinking like someone shipping decisions. Since then, here's what that's looked like in practice: → Built a credit risk model on 150K+ loan records, hitting 0.87 ROC-AUC, with SHAP explainability baked in → Rebuilt SQL pipelines that cut query time by 40%, on datasets with 50K+ records → Turned raw manufacturing telemetry (160K+ records) into dashboards that pinpointed exact causes of downtime I work best with teams that want someone who owns the full pipeline - raw data to deployed insight - not just someone who runs one query and hands off a chart. I'm at my best when I get to ask 'why' behind the numbers, not just report them. If you're hiring for an analyst or early-career data/tech role and want someone who thinks in outcomes, not just outputs, let's talk. Message me, or check the projects below.

Location

Jaipur

AI stack

antigravity

claude code

cursor

gemini

hugging face

openai

Education

Bachelor of Technology - BTech, Computer Science, Government Engineering College, Ajmer

Sophia Secondary School

Public links

github: https://github.com/sakshidatwani