This page separates enterprise delivery, research platforms, and independent builder work so execution is easy to read without digging through a mixed bag.
78-attribute MicroStrategy analytics platform integrating SmartPath, Maximo, IPACT, and LDAP into one decision surface.
Built derived metrics, conditional formatting, cross-filter interactivity, and a structured migration path from development to production with director sign-off.
Three-tier ETL pipeline integrating SmartPath API, Maximo, IPACT, and LDAP into unified Control Plan reporting.
Processes 150,000+ records in roughly 20 minutes across staging, warehouse, and analytics layers using schema-aware loads across DEV, QA, and PROD.
Stateful Python algorithm computing three duration metrics for agent work cycles after fixing event sequencing defects.
Refactored a flawed sequential method into a robust two-pass group-by propagation model using deterministic anchoring and backfill logic.
Full-stack RCA effort that corrected historical data integrity drift and restored analytical confidence.
Executed staged historical recasts correcting 78,000+ records, expanding analytical coverage from 1 month to 9+ months and improving match accuracy to the strongest level since inception.
Performance re-architecture that replaced a direct join to a 23-million-row live table with a pre-aggregated reference design.
Cut runtime from 12 minutes to 2 minutes, eliminated timeouts, and stabilized nightly SLA compliance.
Transformation logic to standardize values across four different source systems using COALESCE waterfall patterns.
Fixed 80,000 product discrepancies and created unified customer name, product, and ticket number fields.
Open-source full-stack bioinformatics platform used by researchers for visualization, simulation, and analysis workflows.
Built with React, D3.js, R Shiny, Python, WebSockets, and Docker using microservices architecture and SOLID design principles.
Scalable processing pipeline integrating DISQOVER, ENCODE, PCAWG, PRIDE, and TCGA data for cancer biomarker analysis.
Applied SVM-RFE, Random Forest, and HPC workflows to help identify 8 novel biomarkers and accelerate validation timelines.
Interactive exploration environment for microbial community data with taxonomic profiling and comparative metagenomics views.
Built with a React frontend and Python backend for 16S rRNA sequencing workflows.
Automated quality and anomaly detection pipelines combining unsupervised ML with rule-based heuristics for large datasets.
Integrated with ETL and CI/CD workflows to flag issues in real time and improve data integrity.
Open-source educational platform for cellular and molecular biology, built as a free alternative to Cell Collective with interactive tutorials, quizzes, and ODE-based pathway simulations.
R/Shiny frontend with a Python/Flask backend that runs SBML simulations through Tellurium and falls back to a built-in R deSolve solver so the app works fully standalone. Eleven pathways shipped with tutorials, quizzes, and interactive SVG diagrams. Four pathways (glycolysis, TCA cycle, DNA replication, protein synthesis) have full ODE simulations with adjustable parameters and real-time Plotly charts. Production stack is Docker Compose with Nginx and gunicorn, and four GitHub Actions pipelines cover R lint and tests, Python lint and tests, Docker smoke checks, and image publishing on version tags.
Live sandbox for product experiments, technical prototypes, and ideas that are still taking shape.
Acts as a public workshop for testing product direction, interaction patterns, and small engineering concepts before they graduate into fuller standalone projects.
Hub for the ProofMark document-craft tool line — one catalog of ~50 PDF, text, and publishing utilities built as a single React SPA over a thin FastAPI shell.
Three sibling FastAPI apps (hub, proofmark-pdf, text-cleaner) composed by URL rather than imports, so each surface stays independently editable, deployable, and testable. The hub renders the catalog, routes to each tool, and shares a design system of color tokens and SVG illustrations across tools.
Public-interest job board that pulls nonprofit tech roles across Canada and the US into one filterable place instead of leaving them scattered.
Automated Greenhouse ingestion for Wikimedia, Khan Academy, and CodePath, with manual placeholders for Code for America, Mozilla, CanadaHelps, Open North, and GiveDirectly so the coverage gap stays visible. Built on Next.js, TypeScript, and Postgres with a seed-data fallback so the local demo still works without a database. Scheduled refresh hits a protected ingestion endpoint hourly via GitHub Actions.
For-profit sibling of Signal for Good. Same product shape, but the source catalog is tuned for corporate hiring across ten ATS families.
Automated adapters for Greenhouse, Lever, Ashby, SmartRecruiters, Recruitee, Teamtailor, Workable, YC Jobs, Wellfound, and generic career pages. Cross-source dedupe runs in the shared merge layer so repeat postings collapse before display. Storage is fully isolated from the Good board through dedicated profit_* tables, so both can share a Postgres instance without mixing nonprofit and for-profit records.
The portfolio itself, rebuilt to turn a content-heavy profile into a guided, high-signal engineering narrative.
Built with Next.js and React as a structured system for showcasing execution, depth, and professional signal without losing factual detail.
The project list shows the systems. The experience archive shows ownership, scope, and how those systems played out in live environments.