I'm looking for enhancement and optimization ideas for a per...

Prompt

I'm looking for enhancement and optimization ideas for a personal software project. Please ignore any constraints I currently work under (listed at the end for reference only) and suggest what you think would make the project most valuable. I'll decide later whether any rule is worth bending. ## What it is "Austin Watch" is a private, single-user, map-first dashboard of the central Austin, Texas housing market (12 ZIP codes: 78701–78705, 78722, 78723, 78731, 78751, 78752, 78756, 78757). Purpose: [one line — e.g. "learn the market and spot good buys before purchasing a home in the next 6–12 months"]. ## How it works today - Python data pipeline run on a schedule by GitHub Actions in a private repo. Each data source is a small adapter; each map layer and each per-property fact is a YAML manifest. Data is stored as Parquet/JSON and committed back to the repo. Every run is logged with calls made, rows, errors and the source's as-of date. - Web app: React + MapLibre GL, served locally on my machine (no public deploy). - Daily listing sweep: a brokerage VOW (Virtual Office Website) listing feed, split into 96 slices (ZIP × property type × price band), refreshed on a rolling basis under a daily call budget. About 1,250 active listings. Tracks price history, status changes, new listings and listings that leave the feed. - Each listing is matched to its county appraisal-district (TCAD) parcel, giving appraised values and 2021–2025 appraisal history. ## Current features - Listing map with filters (price, size, beds/baths, days on market, year built, price per sq ft, asking vs. appraised value, HOA, price cuts), saved filter presets, watchlist, notes, side-by-side compare, and a shareable map bookmark. - Per-listing panel: price history, appraisal history, estimated monthly cost at asking price (2026 property-tax stack, weekly mortgage rate), and facts: FEMA and City floodplains, creek buffers, wildfire risk zone, zoning, city jurisdiction, building permits on the parcel, nearby development cases and permits, assigned AISD schools with TEA ratings, transit access (CapMetro scheduled frequency, planned Project Connect stations), utility providers, nearby parks/trails/libraries/amenities, crime by Census block group, and neighborhood demographics. - Map layers: all of the above as layers, plus ACS 2020–2024 neighborhood estimates (income, home value, rent, age, households, education, commute, vehicles, language, mobility, race/ethnicity) at block-group and tract level, each with margins of error and reliability flags. - Weekly digest page: new listings, price changes, status changes and removals, with per-ZIP counts and medians. - Sources page showing each source's freshness and as-of date. Every number carries its source and date; test data is labeled. - Safeguards: call budgets per source, time limits per stage, a guard that keeps the previous map layer if a refresh comes back empty or sharply smaller. ## Current constraints (for reference — you may ignore them) 1. Private: listing data is never public; the app runs only on my machine or behind a login. 2. $0: no paid services and nothing that needs a credit card on file. GitHub Actions free tier (2,000 minutes/month; current use ~1,800). 3. Polite source use: strict daily call budgets per source (listings 150 requests/day); no bulk scraping. 4. Single user; no accounts, no sharing. 5. Code is written mostly by AI coding agents, reviewed through pull requests with automated tests. ## What I'd like from you Suggest up to 12 enhancements or optimizations: new analyses, data sources, features, architecture changes, or ways to work differently. Favor ideas that would change what I decide or notice, not just add more data. Skip anything already listed above unless you'd change it substantially. For each idea, use this format: - **Idea** (one line) - **Why it matters** (2–3 sentences: what decision or insight it improves) - **How** (data sources, approach, rough build steps) - **Constraint it bends** (numbers from the list above, or "none") - **Effort** (S / M / L) and **ongoing cost** (dollars/month, or $0) - **Risks or caveats** (data licensing or terms of service, accuracy, maintenance) Then: - Rank your top 3 and say why. - Name anything in the current design you would remove or simplify. - If you think one of my constraints is costing me the most value, say which and why.