Case studies & portfolio artifacts
Open-source platforms I run and a growing portfolio of smaller artifacts, data tools, design pieces, experiments, and research outputs.
Pyaar Media
A personal media operating system: curate links and write takes in Notion, publish to a Bluesky account I own. First wing is Pyaar Health News, healthcare and medical-device news for physicians, patients, and researchers.
Letterboxd → Notion: Bridging Feed-less Apps into a Personal Content Database
Using RSS as a universal adapter to mirror my Letterboxd film diary into a Notion database, a reproducible pattern for turning any app into a feed for your own "second brain," even when it has no official API.
PDOS
A Notion-native operating system for the ops manager of a small team: projects, role-by-role workstreams, launch cadence, and a contacts rolodex in one workspace. An optional Bridge layer connects it to WhatsApp, Slack, Discord, or Telegram
pyaar-dicom
A local-first DICOM annotation station for crowdsourcing radiologist validation of medical imaging datasets. Pairs Apple's newly DICOM-approved Studio Display XDR with Kitware's open-source viewer, the public sandbox runs on FDA-released synthetic mammograms; the local install keeps real PHI on the radiologist's own machine.
Pyaar Trivia
An open-source trivia hosting platform built for live pub quiz nights. Full-screen presenter mode, real-time scoring grids, AI-powered question generation via Claude, and support for multiple quiz formats including progressive reveal, video rounds, and picture rounds. Used weekly at venues across Atlanta and Chicago for 3+ years.
SaMD Clearance Visualization
Interactive visualization of FDA 510(k) clearances for Software as a Medical Device (SaMD). Explores regulatory trends, predicate device networks, and clearance timelines across product codes.
FDA CRL Analyzer
Analyzes 202 FDA Complete Response Letters (2020-2024) to extract regulatory intelligence, deficiency patterns, therapeutic area trends, and sentiment. Uses DuckDB for fast SQL analytics and Parquet for efficient storage. 8 deficiency categories, 7 therapeutic areas.
NTS Extractor
A web app that extracts tracklists from NTS Radio show pages. Flask backend scrapes NTS pages and uses bracket-counting JSON extraction to parse tracklist data. Brutalist design aesthetic. Built with Python and vanilla JS.
Logo Letterhead Tool
A web tool for generating branded PDF letterheads with logo placement. Uses pdf-lib for PDF generation, sharp for image processing, and ML-based background removal. Per-page control over logo position and size. Built with Next.js.
LinkedIn Notes
A Chrome extension that adds private notes and color-coded tags to LinkedIn profiles. All data stays local in the browser, no servers, no accounts, no tracking. Includes a CRM dashboard, quick search popup, and JSON export. Built with WXT + React.
NBA Match Dashboard
An interactive NBA game analysis dashboard with D3.js hexbin shot charts, force-directed assist networks, and score timelines. Built with Next.js 15, React 19, and a Python data fetcher using NBA API V3 endpoints.
FDA Devices MCP Server
An MCP server that gives AI assistants structured access to FDA medical device databases, 510(k) clearances, PMA approvals, adverse events, recalls, and device classification. Built with TypeScript, includes synonym expansion and source links.
PYAAR Radio
A personal DJ music app with 41K+ tracks, DuckDB WASM for in-browser SQL filtering, Camelot harmonic mixing, and a daily auto-sync pipeline from YouTube Music. Built with Next.js, React 19, and Tailwind.
Innolitics Medtech Newsletter
Two years writing and growing Innolitics' biweekly Medtech newsletter on FDA regulatory strategy for AI/ML Software as a Medical Device (SaMD). Grew it to 2,500+ subscribers, with 510(k) breakdowns, predicate analysis, AI/ML guidance, and submission lessons for founders.
Drug Name Normalizer
A client-side web app that normalizes messy medication names to standardized generic names using the NIH RxNorm API. Built with Next.js 15, TypeScript, and TailwindCSS. Solves a real healthcare data problem, converting 2-4 hours of manual drug mapping into a 2-minute automated process.
Notion-Based Glucose Dashboard
A visualization of 14 days of my glucose data (or yours if you want). Now includes Part 2: an Abbott Lingo vs Dexcom Stelo comparison, built with a Python pipeline and explored via a Notion MCP server.