Projects

Applied analytics, dashboards, research prototypes, and data systems.

A recruiter-focused project gallery with filters, case-study links, verified demos, and clear prototype labels.

Filterable gallery

Explore the work by role relevance.

Showing all projects.

Search projects

Showing all projects.

business-intelligencehealthcare-analyticstableau

Lead Intelligence and Scoring Dashboard

Problem

The organization needed a clearer way to identify high-priority healthcare leads, monitor engagement, and support outreach decisions.

Context

Applied analytics project framed as a stakeholder decision-support engagement, using public-safe or portfolio-safe materials for recruiter review.

Method

Cleaned and standardized lead records; created KPI scorecards; analyzed lead source, owner, engagement, contact completeness, and readiness; built a Top-50 prioritization queue with filters and drill-downs.

Tools

Tableau · Excel · Google Sheets · SQL · Python · Pandas

Outcome

Improved visibility into lead quality and supported faster outreach prioritization.

Business Impact

Improved visibility into lead quality and gave stakeholders a clearer way to prioritize outreach, monitor readiness, and discuss pipeline quality.

Technical Challenges

Required standardizing fragmented lead fields, defining transparent scoring logic, and presenting contact readiness without exposing confidential records.

Lessons Learned

This project strengthened my ability to transform fragmented healthcare lead data into an executive-ready prioritization dashboard while balancing technical accuracy with business usability.

Repository: available on request · Live demo available

Live Demo / Case Study Related research on business intelligence visualization

business-intelligencetableau

Competitor Pricing and Market Intelligence

Problem

Medicidiom needed a structured view of competitor pricing, product features, market positioning, and potential strategic gaps.

Context

Applied analytics project framed as a stakeholder decision-support engagement, using public-safe or portfolio-safe materials for recruiter review.

Method

Benchmarked 14 medical-language and healthcare-related competitors across pricing models, AI capability, delivery models, geographic focus, certification, and B2B positioning.

Tools

Excel · Google Sheets · Tableau · Competitive analysis · Market intelligence

Outcome

Identified priority threats and market gaps and produced an internal decision-support resource.

Business Impact

Turned scattered competitor information into a structured view of pricing, feature positioning, and market gaps for strategy conversations.

Technical Challenges

Required normalizing inconsistent public competitor details and separating evidence-backed insights from assumptions.

Lessons Learned

This project reinforced the value of turning scattered market signals into a structured recommendation framework that leaders can use without needing to inspect every raw data point.

Repository: available on request · Live demo available

View competitor market intelligence case study

business-intelligencetableau

LinkedIn Performance Analytics Dashboard

Problem

The team needed visibility into social performance, audience composition, and content engagement.

Context

Applied analytics project framed as a stakeholder decision-support engagement, using public-safe or portfolio-safe materials for recruiter review.

Method

Analyzed impressions, reactions, engagement patterns, geography, job functions, seniority, industries, and post performance.

Tools

Tableau · Excel / Google Sheets · KPI development · Data visualization

Outcome

Created an interactive reporting view for content and audience analysis.

Business Impact

Connected content performance with audience composition so stakeholders could understand reach, engagement, and growth opportunities.

Technical Challenges

Required balancing time-series trends, audience segmentation, and dashboard density without overloading non-technical users.

Lessons Learned

This project sharpened my ability to connect content metrics, audience segments, and growth decisions in a way that supports clearer communication strategy.

Repository: available on request · Live demo available

Live Demo / Case Study

healthcare-analyticsmachine-learningsql-data-engineering

Medical Appointment No-Show Analysis

Problem

Identify patterns associated with missed healthcare appointments and possible opportunities for outreach improvement.

Context

Applied analytics project framed as a stakeholder decision-support engagement, using public-safe or portfolio-safe materials for recruiter review.

Method

Cleaned and explored more than 100,000 appointment records; examined age, SMS reminders, scholarship status, weekday, and related features; translated findings into operational recommendations.

Tools

Python · Pandas · Matplotlib · Scikit-learn

Outcome

Identified behavioral and scheduling patterns that can support appointment outreach and resource planning.

Business Impact

Translated appointment data into outreach and scheduling insights that could inform reminder strategy, patient communication, and operational planning.

Technical Challenges

Required careful cleaning, feature interpretation, and clear separation between correlation, prediction, and clinical claims.

Lessons Learned

This project strengthened my ability to move from exploratory healthcare data analysis to practical outreach recommendations while clearly communicating limitations.

Repository: available on request · Live demo available

View healthcare no-show analysis case study

healthcare-analyticssql-data-engineeringtableau

Healthcare Data Pipeline and KPI Reporting

Problem

Fragmented CSV and claims-style datasets made consistent reporting difficult.

Context

Applied analytics project framed as a stakeholder decision-support engagement, using public-safe or portfolio-safe materials for recruiter review.

Method

Combined multiple files and reporting inputs; used SQL and Python for cleaning, transformation, and validation; built weekly KPI scorecards and prepared data for Tableau reporting.

Tools

SQL · Python · Pandas · Excel · Tableau

Outcome

Improved repeatability and consistency of reporting while reducing manual preparation.

Business Impact

Improved repeatability of KPI reporting by showing how raw healthcare files can be cleaned, validated, and prepared for dashboard use.

Technical Challenges

Required designing transformations that are transparent, reproducible, and easy to validate before visualization.

Lessons Learned

This project reinforced the importance of repeatable data preparation, validation, and documentation before any dashboard or KPI report can be trusted.

Repository: available on request · Demo not currently available

healthcare-analyticsmachine-learningresearch-prototypes

Federated Deep Learning for Multi-Hospital Clinical Prediction

Problem

Train predictive models across institutions without centralizing sensitive patient data.

Context

Applied analytics project framed as a stakeholder decision-support engagement, using public-safe or portfolio-safe materials for recruiter review.

Method

Prototyped simulated hospital nodes, trained a shared model without raw data exchange, evaluated privacy-utility trade-offs, and explored differential privacy noise injection.

Tools

Python · PyTorch · Federated learning · Differential privacy

Outcome

Demonstrated a privacy-preserving approach to collaborative clinical prediction. Research prototype only.

Business Impact

Demonstrated how institutions could collaborate on predictive modeling while reducing the need to centralize sensitive clinical records.

Technical Challenges

Required simulating distributed nodes, reasoning about privacy-utility tradeoffs, and clearly labeling the work as a research prototype.

Lessons Learned

This project deepened my understanding of privacy-preserving model development and the trade-offs between collaboration, utility, and data protection.

Repository: available on request · Demo not currently available

healthcare-analyticsmachine-learningresearch-prototypes

LLM Agent for Clinical Documentation Support

Problem

Explore whether a retrieval-augmented AI assistant can support clinical-documentation workflows while reducing hallucination risk.

Context

Applied analytics project framed as a stakeholder decision-support engagement, using public-safe or portfolio-safe materials for recruiter review.

Method

Built a retrieval-augmented, tool-using LLM workflow, connected structured lookup tools, generated clinical-note summaries, and evaluated outputs against reference material.

Tools

LangChain · OpenAI API · RAG · Python · Tool-using agents

Outcome

Demonstrated a prototype workflow for grounded documentation support. Prototype only; not a clinically validated system.

Business Impact

Showed how grounded AI assistance could support documentation workflows while keeping risk controls and human review central to the design.

Technical Challenges

Required retrieval grounding, tool-use design, and evaluation thinking to reduce hallucination risk and communicate workflow limitations.

Lessons Learned

This project taught me to design AI prototypes around grounding, evaluation, and risk controls rather than treating generative output as automatically reliable.

Repository: available on request · Demo not currently available

machine-learning

Student Admission Prediction

Problem

Model educational admission outcomes with transparent evaluation and limitations.

Context

Applied analytics project framed as a stakeholder decision-support engagement, using public-safe or portfolio-safe materials for recruiter review.

Method

Built a logistic-regression pipeline and evaluated performance with accuracy, ROC-AUC, precision, recall, F1, and confusion matrices.

Tools

Python · Scikit-learn · Logistic regression · Model evaluation

Outcome

Educational machine-learning project; performance should be interpreted with dataset limitations and leakage checks.

Business Impact

Provided an educational example of transparent predictive modeling and evaluation for admissions-style decision support.

Technical Challenges

Required checking modeling assumptions, evaluation metrics, potential data leakage, and fairness limitations.

Lessons Learned

This project reinforced the need for transparent evaluation, leakage checks, and careful interpretation when building predictive models for education use cases.

Repository: available on request · Demo not currently available

machine-learninghealthcare-analytics

Heart Disease Prediction

Problem

Explore classification workflows on a heart-disease dataset.

Context

Applied analytics project framed as a stakeholder decision-support engagement, using public-safe or portfolio-safe materials for recruiter review.

Method

Built a KNN classification pipeline and evaluated performance with accuracy, ROC-AUC, precision, recall, F1, and confusion matrices.

Tools

Python · Scikit-learn · KNN · Model evaluation

Outcome

Educational machine-learning project, not a clinical diagnostic tool.

Business Impact

Demonstrated a healthcare classification workflow while making clear that the work is educational and not a diagnostic system.

Technical Challenges

Required careful feature handling, metric interpretation, and responsible communication of clinical limitations.

Lessons Learned

This project strengthened my understanding of healthcare model limitations and why educational classifiers should never be presented as clinical diagnostic tools.

Repository: available on request · Demo not currently available

healthcare-analyticssql-data-engineering

MyClinic Secure Healthcare Cloud Architecture

Problem

Design a scalable and secure serverless architecture for a healthcare patient portal.

Context

Applied analytics project framed as a stakeholder decision-support engagement, using public-safe or portfolio-safe materials for recruiter review.

Method

Mapped access, frontend, API, compute, storage, identity, encryption, WAF, and governance controls into a clear healthcare architecture diagram.

Tools

AWS Route 53 · CloudFront · S3 · API Gateway · Lambda · DynamoDB · Cognito · KMS · WAF · IAM

Outcome

Architecture project focused on security, scalability, resilience, and patient-data protection.

Business Impact

Converted a healthcare portal concept into a security-aware cloud architecture story that reviewers can understand quickly.

Technical Challenges

Required mapping AWS services into a coherent flow across identity, API, compute, storage, encryption, and edge protection.

Lessons Learned

This project helped me connect cloud service selection, security controls, and architecture storytelling into a diagram that both technical and non-technical reviewers can follow.

Repository: available on request · Live demo available

Live Demo / Case Study Related publication: Resilient Dataflow Intelligence