Jeff Kirk Jr logo
Self-hosted AI & Data Systems · Data Analytics · Investment Operations
Portrait of Jeff "Jay" Kirk Jr.

I'm Jeff "Jay" Kirk Jr. — I spent eleven years reconciling other people's numbers. Now I build systems that reconcile their own.

11+ years in institutional investment operations (Man Group, State Street), supporting post-trade operations for portfolios up to $45B AUM — now building a trading scanner, a statewide property engine, and a market-gap mapper that run daily on live data. Self-hosted, cron-driven, and governed by one rule: a figure that can't survive its own recomputation doesn't ship.

Featured Projects

Rental Intelligence Engine

305,476 Buildings in DB $290B Assessed 14 Counties · MA Interactive Map
Problem

Everything that signals a good multifamily deal is public — parcel geometry, assessed values, owner names, HUD rents — but it sits scattered across town assessor databases, a statewide GIS layer, and federal rent tables nobody joins. Screening Massachusetts' 305,476 multifamily buildings town-by-town by hand never finishes, and it's blind to the owner-level patterns that actually flag a deal.

What I built

A statewide multifamily acquisition research platform covering all 305,476 multifamily buildings across Massachusetts. It analyzes properties by cap rate, NOI, and ownership signals, and identifies portfolio owners and acquisition targets.

Tech stack

MassGIS parcel data, HUD Fair Market Rent benchmarks, Python data pipeline, interactive web map.

Outcome

305,476 buildings in the database across $290B in assessed value; 14,982 carry active opportunity scores (0–100) — cap rate, NOI, and ownership signals surfaced for any Massachusetts submarket in seconds.

Trading Intelligence Dashboard

2,212 Symbols 12 Scanners DCF Valuations Live Since 2025
Problem

A human can scan 2,212 symbols for about a week — then attention drifts, the routine slips, and worst of all there's no honest record of whether the picks were ever any good. The failure mode isn't missing a setup; it's keeping no score.

What I built

A self-hosted trading-research platform that scans a 2,212-symbol universe daily with 12 automated scanners, regime detection, EDGAR DCF valuations, conviction scoring, and daily Polygon.io price updates. Delivers high-conviction picks via Telegram alerts, behind secure authentication.

Tech stack

Python, Polygon.io & Finnhub APIs, SEC EDGAR filings, SQLite, cron-scheduled pipeline, Telegram alerts, authenticated web dashboard.

Outcome

12 scanners run daily across a 2,212-symbol universe, delivering high-conviction picks to Telegram with a live track record on the dashboard.

Business Intelligence Engine

334,638 Businesses Sector-gap analysis · OpenStreetMap 885 ZIP Zones ZIP · Town · County
Problem

A market gap is an absence, and absences don't show up in a directory — they only exist by comparing a place against its peers, across many sectors at once. Doing that by eye over 334,638 Massachusetts businesses in 885 ZIP-level zones doesn't fail loudly; it just quietly never finishes.

What I built

Spatial analysis engine that surfaces geographic areas with economic imbalances, underserved sectors, and unrealized business potential. Analyzes industry density, diversity, and sector gaps across Massachusetts ZIP codes, towns, and counties.

Tech stack

OpenStreetMap (OSM) place data, spatial/geographic analysis, Python, interactive map interface.

Outcome

334,638 businesses mapped across 885 ZIP zones — sector density and coverage gaps surfaced for every Massachusetts town and county via interactive map.

Other Projects

Dev Tools & AI Infrastructure

AI Infrastructure Home Lab CLI Tools Self-Hosted
What I built

Self-hosted developer utilities and AI infrastructure. Tools include WiFi scanner, USB recovery toolkit, system diagnostics, duplicate detector, and web scraping utilities — built and run on my own machines. Home lab runs local AI agent workflows and memory systems across networked machines.

Tech stack

Python, Bash, self-hosted Linux services, local AI agent workflows.

Explore →