Open to data analyst roles
Finance · Healthcare · Consulting analytics

Hi, I'm Akansha Singh

A final-year student with a biotech mind and an analyst's toolkit — I turn messy, real-world data into decisions people can act on. Ten shipped projects, from warehouses and risk models to a fraud-detection audit engine.

10
Projects shipped
900K+
Records analysed
3
Analytics domains
2+
Years in research
// about

Start before you're ready

I'm a Biotechnology student at Amity University, Noida who believes the best way to learn something is to start before you feel ready.

That's how I ended up interning at IIT, competing at national hackathons, and spending two years in a research lab. Data analysis is where that instinct meets a real skill — I'm drawn to the moment when numbers stop being numbers and start telling you something true about a patient, a business, or a decision that needs to be made. I'm working towards roles in finance, healthcare, and consulting analytics.

Vienna waits for you — so there's time to build something worth arriving at.

EducationB.Tech Biotechnology · Amity University · CGPA 8.49
CurrentlyResearch Intern · IIT (ISM) Dhanbad · SRIS 2026
ExperienceData Analyst Intern · Bluestock™ · Jun 2026
ResearchSystems Biology & Data Analytics Lab · Amity Institute of Biotechnology · 2+ years
TrainingCSIR-CIMFR · Air-pollution-tolerant microbes from coal-mining areas · Python analysis · Jul 2026
LeadershipJoint Secretary · Innovation & Entrepreneurship Cell, AIB
Finalist · Riverathon 1.0Ministry of Jal Shakti, NMCG & World Bank
Semi-finalist · IIT Delhi Handloom Hackathon300+ teams · Ministry of Textiles
10th 95.2% · 12th 93.4%Delhi Public School
// toolkit

The stack behind the work

Languages

PythonSQLGitExcel

Data & Analytics

PandasNumPyscikit-learnSHAPMatplotlibSeabornETL PipelinesData Warehousing

Deployment

FlaskStreamlitOpenCV

Focus Areas

Healthcare AnalyticsFinancial AnalyticsConsulting

Methods

ClassificationHypothesis TestingAnomaly DetectionFeature EngineeringCross-ValidationRAG / LLMs

Domains

BioinformaticsComputer VisionEDAData Storytelling
// selected work

Projects that ship a decision, not just a chart

Every project below is a real dataset, a real question, and a result someone could act on. Full code, data, and write-ups on GitHub.

📈

Mutual Fund Analytics Warehouse

Flagship

An end-to-end analytics warehouse over 100K+ rows of AMFI-style mutual-fund data — a MySQL star schema, a tested Python ETL pipeline, automated data-quality gates, and an Excel dashboard that turns ten raw CSV feeds into board-ready industry metrics.

102 automated tests 58 data-quality checks Direct 0.78% vs Regular 1.35% TER
MySQLPython ETLpytestExcel DashboardFinancial Analytics
View on GitHub
💳

Credit Default Risk Analytics

Flagship

Predicts credit-card default on 30,000 real clients and turns the model into a money decision — imbalanced classification judged on precision-recall, a cost-based approve/decline threshold, and SHAP explainability that a credit officer (and a regulator) can actually act on. The core analytics job at a card issuer.

PR-AUC 0.56 vs 0.22 baseline Cost-optimal threshold cuts loss 26% Catches 75% of defaulters
MySQLscikit-learnSHAPImbalanced ClassificationFinancial Analytics
View on GitHub
🩺

30-Day Readmission Risk Model

Flagship

Predicts 30-day hospital readmission on 101,766 real patient encounters — SQL cohort analysis, hypothesis testing, and a risk model that ranks patients before discharge so care teams can intervene where it actually matters.

ROC-AUC 0.64 Top 30% captures 48% of readmissions 101,766 encounters
MySQLPythonscikit-learnHypothesis TestingHealthcare
View on GitHub
💬

FundGPT — RAG over Fund Factsheets

Flagship

Ask "which fund has the lowest expense ratio?" and get a cited answer. A retrieval-augmented Q&A engine over 40 factsheets generated from the warehouse above — a zero-cost structured answer engine, with optional Claude synthesis for open-ended questions.

100% retrieval hit-rate@1 Comparison tables & rankings Cited by AMFI code
PythonRAGTF-IDFStreamlitGenAI
View on GitHub
🔍

Healthcare Claims FWA — Payer Audit Prioritization

Flagship

Ranks 200+ NY hospitals by billing-outlier risk across 111,155 real claims lines and screens the provider network against the 83,464-entry federal exclusion list — the payment-integrity workflow at Optum, a Blue plan, or CMS. Two independent detectors (peer-benchmark markup + Isolation Forest) have to agree before a facility hits the priority audit queue, cutting false positives.

6 of 200 flagged by both detectors 111K billing lines · 204 hospitals 83K-provider exclusion screen
MySQLscikit-learnIsolation ForestRobust StatisticsHealthcare
View on GitHub
More projects
🏥

Hospital Readmissions Risk Analysis

Analysed 64,764 CMS records across 4,239 hospitals to find which conditions, states, and hospitals drive 30-day readmission risk — and whether size or prestige actually predicts performance.

15-pt protocol gap (16.3%–31.3% HF) · Excel dashboard · exec PDF
MySQLPythonCMS DataHealthcare
GitHub
📉

Equity Settlement Risk Analysis

Analysed 507,521 real SEC settlement-failure records to locate where operational risk concentrates, when stress events hit, and how to tell chronic failures from event-driven spikes.

$147B March stress event · 18% top-10 concentration · 7 SQL queries
MySQLPythonSEC DataRisk Analytics
GitHub
💊

Medical Cost Forecasting

Forecast 10-year US healthcare spending across 6 CMS service lines — benchmarking Prophet, ARIMA, and Holt-Winters on a 5-year holdout to find which model wins where, and why.

No single model dominates · COVID modelled as event shock · Excel dashboard
ProphetTime SeriesMySQLPython
GitHub
🍄

Fungal Pathogenicity Predictor

Classifies fungal disease risk across 9 pathogen types from ecological trait data — applicable to pharma R&D, agricultural risk, and life-sciences analytics.

92.2% accuracy · 0.683 macro F1 · 9 classes
Gradient BoostingSHAPsklearn5-Fold CV
GitHub
📊

Student Performance Predictor

An end-to-end Flask app that predicts academic outcomes and generates a Student Success Index with personalised recommendations — the same pattern as patient risk scoring and churn.

End-to-end · Flask dashboard · actionable insights
Flaskscikit-learnPrediction
GitHub
// let's talk

Open to data analyst roles

Finance, healthcare, or consulting — if you're working on something data-driven and want someone who pairs domain curiosity with analytical rigour, let's talk.