HELLO, I'M NIHIRA SHARMA
Advanced Computing student at the University of Sydney, building risk models, decision systems, and interactive analytics products end-to-end - from SQL pipelines to deployed dashboards.
Selected Work · 06
Six end-to-end projects across credit risk, fraud, quant volatility, churn, experimentation, and data tooling. Each one goes from raw data to a deployed decision surface.
Interpretable credit scorecard on 307K applicants - WOE/IV, PDO scaling, fairness & calibration validation, deployed as a live 4-page Shiny app.
↳ 58M+ source rows aggregated across 7 relational tables
SQL-first feature engineering on 1.3M transactions; XGBoost lifted precision from 17% to 99.2% at equal recall, with an expected-loss review queue.
↳ Validated on the synthetic Sparkov dataset
GARCH family volatility models (sGARCH / eGARCH / gjrGARCH) with VaR, CVaR, and Kupiec backtesting - deployed as a live multi-ticker Shiny app.
↳ Live-data GARCH comparison with Kupiec backtests
Churn model with SHAP explanations, four customer segments, and a revenue-at-risk workflow - leakage identified and removed before final scoring.
Frequentist + Bayesian experimentation engine with SRM checks and ship/no-ship recommendations - validated on 294K real e-commerce observations, with automated PDF reporting.
Upload-any-CSV workflow that returns quality warnings, outliers, skewness, correlations and target analysis - deployed as a public Streamlit app.
ATS-scoring and skills-extraction engine using the ESCO taxonomy and live Adzuna job data. Finishing document parsing and NER extraction now.
Why me
A quick read of the evidence - what I've actually shipped, not skills I've just listed.
End-to-end builder
I don't hand off. Every project owns the SQL pipeline, the modelling, the validation, and the interface a stakeholder actually uses.
Documented projects
Rows engineered · PostgreSQL · Credit Risk
Fraud PR-AUC · Sparkov synthetic
Data + business
Technical analysis wired to the decision it changes.
SQL-first
Materialized views, window functions, 7-table joins before the model runs.
USYD
B. Advanced Computing · Data Science + Business Analytics · June 2027.
About · nihira@portfolio
I'm Nihira Sharma, an Advanced Computing student at the University of Sydney based in Sydney, NSW.
I build risk, decision, and analytics products end-to-end - SQL pipelines to deployed dashboards - and care most about work that holds up out-of-sample.
Right now I'm open to 2026–27 internships and graduate roles in data science, risk analytics, ML engineering and decision intelligence.
Education & Leadership
University of Sydney
Coursework: Data Structures & Algorithms, Intro to AI, Predictive Analytics, Business Forecasting, Statistical Modelling for Business, Scalable Data Management (PySpark/Databricks), Data & Information Management.
USYD · Industry-linked
ARMAX and HAR-RV volatility modelling on Optiver market data.
“Accelerating AI Adoption for India's Largest Online Grocer”
Applied AI adoption strategy and analytics for a large-scale grocery platform.
University of Sydney
Supporting new students through academic transition and course navigation.
Data science · finance · computing
Took the lead across multiple group assignments - scoping, dividing work, keeping delivery on track.
Let's build
Open to internship and graduate roles in data science, risk analytics, ML and decision intelligence. Sydney-based, remote-friendly.
© 2026 · Nihira Sharma · Sydney, NSW
Available · Graduating June 2027