• Currently: Working with PrimeIQ.ai
• Interests: Baseball & sports analytics
• Education: Boston University
• Location: San Francisco Bay Area
• Connect: linkedin.com/in/harlanljones

Active cohort on MLB active rosters today (6): Paul Goldschmidt (NYY), JJ Wetherholt (STL), Blake Perkins (COL), Landen Roupp (SF), Ray Kerr (ATL), Agustín Ramírez (MIA)

Historical index contains 68 total Major League Baseball players born on September 10.

• urban-signal: Expanded real-time spatio-temporal telemetry streams across 2+ major metros (Corpus Christi, Milwaukee) and built dynamic cross-region comparison analytics.
• dotfiles-showcase: Engineered agent Skills Hub + Git Core & Security cards (HJ-699) and implemented config search index generator with deploy drift check.
• omarchy-agents: Built token usage analytics, designed productivity comparison views and administrative quota limits portals, and refined AI agent monitoring dashboards.

• urban-signal — Real-time spatio-temporal forecasting engine using Apache Kafka, Uber H3 grids, and ST-GNN/LightGBM ensembles to predict hyper-local property appreciation and commercial catalysts ahead of market comps.

• herdr-outpost — Lightweight, secure remote dashboard and relay gateway for Herdr AI coding agent workspaces, featuring mTLS encryption, VAPID authentication, and Cloudflare Workers integration.

• scheme-db — Interactive NFL scheme engineering workstation that animates 80 plays across 20 offensive & defensive systems with Catmull-Rom spline interpolation, 0.0s–1.2s disguise mesh analysis, and coaching lineage trees.

• omarchy-agents — Dashboards and bar widgets for tracking AI coding-agent usage on Omarchy, with a local-first Bun + Hono + React + SQLite web console plus QML bar widgets ranking token usage across local agents.

• clify — Metric-Driven Agent-Driven Development (ADD) framework implementing a Turborepo monorepo with two complementary products: cliamp-clify (Go TUI music player with Spotify superpowers) and clify (Python CLI with ADD-orchestrated agent framework). Features deterministic SLAs on every execution (latency ≤ 2.5s, cost ≤ $0.02, confidence ≥ 0.90), semantic scope guardrails, and runtime telemetry monitoring with three-stage TDD lifecycle verification.

• arbkit — Cross-venue sports & prediction market arbitrage detector and execution simulator in pure Rust, featuring a zero-allocation lock-free SPSC hot loop (p99 < 250ns) ingesting Kalshi & Polymarket feeds in real-time.

• bayes-horizon — S&P 500 long-term ML projection engine using Bayesian hierarchical modeling, walk-forward cross-validation, and point-in-time macroeconomic data (FRB, BLS, CBO, Shiller).

• baseball-dashboard — Live MLB scoreboard and game-day research dashboard with sabermetric matchup stats, ballpark weather, bullpen workload, and optional player props.

• seismic-ml — Spatiotemporal ML pipeline for short-term earthquake forecasting, predicting M ≥ 4.5 events within a 7-day window from 30-day seismic precursor features. Trains a Keras MLP on the USGS catalog (83k+ events) with strict walk-forward splits and leakage-asserting tests, tuned for an RTX 4070 Ti via mixed precision and XLA.

• march-madness — PyTorch deep learning tournament prediction and Monte Carlo bracket simulation engine (BracketNet) trained on 23 matchup differential features and Kaggle historical data.

• aegis — AEGIS‑Thermal: agent-based co‑optimization of battery (BESS) & thermal (TES) storage. Recent work includes g1.5 end-to-end integration demo (negotiation→mandate→dispatch→sensed actuation→receipt) + archived record and w1.4: telemetry baseline demo on live sim (5 families, snapshot TSV) + fault-injection schedule doc (G1.4), built with Rust.

• dotfiles-showcase — Local-first web app to visualize and explore chezmoi dotfiles with live config previews and mini-demos. Recent work includes detailed fallback content strategy and sanitization rules and comprehensive ANSI and theme handling, built with TypeScript and CSS3.

• linear-bucket — Sync and group Linear issues into configurable “buckets” for reporting and workflow automation. Built with Python.

• dotfiles — The dotfiles I use. Recent work includes settings and keymaps for better git integration and separate template execution and apply steps, built with Go Templates and Shell.

• gan-enhanced-pneumonia-classifier — Enhancing Pneumonia Detection with GAN-Generated Synthetic Chest X-rays. Recent work includes new Grad-CAM visualizations for real and synthetic images, enhancing analysis results with additional insights, built with Python.

• ferrite-db — An embedded Rust library for low-latency approximate nearest-neighbor search over dense vector embeddings. Synchronous, blocking API with library-owned parallelism; designed for direct in-process integration without n... Recent work includes wASM bindings for Ferrite DB's in-memory vector engine and reproducible synthetic corpus generator for Ferrite DB benchmark, built with Rust and TypeScript.

• chacha20-avx2 — Active development focused on constant-time tail/scalar fallback path and ratify D1 — Poly1305 accumulator radix = 2^64 (HJ-319), built with C.

• film-lab — Active development focused on deploy verification and analytics and cI/CD for production and PR previews, built with TypeScript and JavaScript.

• web-backends — Active development focused on benchmark infrastructure and Go + Gin reference implementation, built with Shell and Go.