Data Scientist • AI Fairness & Governance • Healthcare Analytics Platform Builder
Amit builds trustworthy, transparent data systems that prioritize fairness in machine learning. Master of Data Analytics student @ UNFC | Platform & Operations Lead @ Synod Intellicare | 4.0 GPA (President's List)
- AI Fairness & Bias Detection – Implementing industry-standard fairness metrics in production systems
- Healthcare AI Governance – Building compliance-ready clinical decision support with bias auditing across demographic subgroups
- Data-Driven Platforms – End-to-end: Python pipelines → Flask APIs → Docker → AWS → CI/CD automation
- Agile Data Science – Product Owner role managing analytics workflows with Scrum methodology
Python • Streamlit • Agile (Scrum)
- Product Owner managing 16 user stories across 2 disciplined sprints
- Predictive analytics dashboard for crime trend analysis
- Full APA technical report and peer/self-evaluation
- TDD-driven development with continuous refactoring
Repo: Toronto_Crime_Indicators_Analysis
Python • pandas • Matplotlib • pytest • Agile (Scrum) • Taiga
A modular Python package that ingests a health-commodity shipment dataset (ARVs and HIV rapid test kits) and turns it into reusable summary functions. Amit led the build as Product Owner across a one-week Agile sprint with a 4-person team.
- Installable
shipment_toolpackage: loader plus country, vendor, and product-group summary modules - Data cleaning for real-world messiness: missing values, non-numeric text in cost and weight columns
- 100% test coverage across the suite (pytest + pytest-cov), with flake8 and black enforced
- Branch-per-feature workflow: every user story merged via peer-reviewed pull request
- Functions return ready-to-use data and figure objects for a future dashboard layer
Repo: supply-chain-shipment-tool (private, course requirement)
React • TypeScript • Express • FastAPI • SQLite • Statistical Analysis
A full-stack analytics platform for Emergency Department (ED) performance, built for the DAMO-699 capstone at the University of Niagara Falls Canada. Two services — a React/Vite frontend behind an Express server (port 3000) and a FastAPI analytics backend (port 8000, JWT-secured, 33 routes) — ingest Canadian Institute for Health Information (CIHI) emergency department data and answer one operational question: what drives ED length of stay, and which patient cohorts are most affected?
- Five pre-registered hypotheses (H1–H5) tested with non-parametric methods suited to skewed clinical data
- Executive dashboard with an Emergency Resource Burden Index (ERBI) and ranked clinical problems
- Throughput forecasting via exponential smoothing with Mann-Kendall trend detection
- Post-cleaning overfitting/underfitting diagnostics with learning and complexity curves
- Multi-format exportable clinical reports, including genuine vector PDF output
- 299 tests across 20 suites (pytest)
Repo: Capstone_Project-DAMO-6994 (private while in development)
Python • FastAPI • SQLite • Leaflet • Docker
A live multi-hazard flood early-warning console for Nepal. It scrapes the country's official hydrological, disaster, and news sources every 12 minutes, scores all 309 DHM river gauges on a 0–100 Flood Severity Index, and forecasts each gauge's trajectory 12 hours ahead.
- Probability of breaching the danger mark within 6 hours, plus a ranked, lead-time-gated action list per gauge
- Hazard layers beyond river thresholds: earthquakes (landslide triggers), active fires, official incidents, and 5-source news coverage
- Models landslide- and moraine-dammed lake outburst floods — the event class behind the July 2025 Rasuwa disaster
- Interactive map with per-river forecast charts, the nearest of Nepal's 16,295 health facilities, verified emergency numbers, and live SSE-pushed updates
- CI and CodeQL on every push; 5 of 6 data sources require no API key
Repo: nepal-flood-watch
Python • Flask • AWS • Docker • GitHub Actions • SQL (In Development)
Production-grade fairness metrics system for clinical ML models:
- Real-time bias detection dashboard (Plotly.js)
- Demographic Parity, Disparate Impact, Equalized Odds
- Automated data validation pipelines
- AWS CodeDeploy integration for zero-downtime deployments
Data access is hand-written parameterized SQL against a versioned
schema.sql. There is no ORM layer, by design, so query plans and index behaviour stay explicit and auditable.
His current work is weighted toward classical statistical inference rather than deep learning: ANOVA, chi-square, Kruskal–Wallis, and OLS regression with assumption diagnostics and model validation.
Amit's contribution patterns, language mix, and consistency at a glance.
- Activity – Total commits, pull requests, and issues across public repositories
- Languages – Primary focus on Python and Jupyter Notebook for analytics work, with TypeScript/HTML for dashboards and web delivery
- Consistency – Ongoing contribution streaks rather than one-off bursts
- Collaboration – The mix of commits and PRs reflects both solo builds and team workflows
Bias in machine learning isn't theoretical. It directly impacts real lives in healthcare, lending, hiring, and criminal justice. Amit builds systems that make fairness measurable, auditable, and enforceable, on the principle that every ML model should ship with fairness metrics the same way it ships with accuracy.
| Category | Skills |
|---|---|
| Machine Learning | Classification, Regression, PCA, Fairness Metrics, Feature Engineering, Model Evaluation |
| Data Engineering | ETL Pipelines, Data Validation, Synthea Simulation, Data Quality Assurance |
| Analytics | EDA, Statistical Analysis, Hypothesis Testing, Dashboard Design |
| Software Engineering | API Development (Flask), CI/CD, Cloud Architecture (AWS), Containerization |
| Healthcare AI | Clinical Data Analysis, Bias Auditing, Compliance Monitoring, HIPAA Fundamentals |
| Agile | Product Owner, Scrum, User Story Writing, Sprint Planning, Taiga |
- Advanced causal inference & counterfactual fairness
- LLM-based data pipelines and prompt engineering
- Cybersecurity in data systems
- Production ML monitoring & observability
Master of Data Analytics (MDA)
University of Niagara Falls Canada | In Progress
President's Academic Distinction List (4.0 GPA)
B.E. Electronics & Communication Engineering
Completed 2020
5+ Years IT Experience (Prior to MDA)
Systems administration, infrastructure, cloud fundamentals
Open to:
- Fairness-focused ML projects
- Healthcare AI governance
- Data science collaboration
- Mentoring on Python & analytics
- Open-source advocate & passionate about ethical AI
- 5+ years in IT before transitioning to data science
Built with ❤️ • Data scientist focused on fairness, governance, and impact