Quantitative developer, systems builder, independent operator.
I design trading infrastructure, build tools that solve real problems,
and think in structures that connect markets, software, and ideas.
I'm a quantitative developer and systems builder based in Houston, TX. My work sits at the intersection of algorithmic trading, software engineering, and applied mathematical reasoning — not as separate disciplines, but as different expressions of the same underlying practice: designing systems that behave correctly under conditions you can’t fully predict in advance.
The primary output of that work is a proprietary trading system I maintain and develop solo — a research, execution, and intelligence environment I’ve been building toward something closer to a one-man quant house than a single strategy. Around that core, I build and release open-source tools for problems I’ve hit personally: multi-provider LLM transcript management, offline training applications, browser automation, AI workflow infrastructure.
The alias Soltfiish carries both. It’s the identity attached to everything I ship — from systems that never leave my environment to tools I release publicly because a solved problem is worth sharing. The through-line isn’t a category of software. It’s a standard of thinking about what I’m building and why.
End-to-end design for algorithmic trading environments — signal modeling, execution logic, multi-strategy coordination, regime awareness, and post-trade analytics.
Building the compute backbone, API integrations, CLI tools, and browser automation that make complex workflows actually run — including always-on homelab infrastructure for AI and trading workloads.
Multi-agent pipelines, MCP protocol, and LLM-powered development workflows — using AI as a force multiplier across research, coding, and system design rather than as a surface-level assistant.
Bounded architecture, drift auditing, and validation discipline applied to systems that must behave predictably as they evolve — especially where AI-generated code and live execution are both in the loop.
A growing panel of public tools and open-source work. Proprietary systems live here too — not open, but worth knowing exist. Each entry is something that reached the bar for being real.
A self-built quantitative research, intelligence, and execution environment — designed and maintained entirely solo. Solt is the operating layer for an integrated trading workflow: signal research, backtesting, execution, performance tracking, and strategy iteration all run through a single coherent system. The longer-range architecture is oriented toward managing multiple accounts, instruments, and market regimes from a unified intelligence stack. It is not a plugin or a bot. It is closer to an institution.
A unified staff training platform covering all seven Darden Restaurants brands — LongHorn Steakhouse, Cheddar's Scratch Kitchen, Yard House, The Capital Grille, Seasons 52, Bahama Breeze, and Eddie V's Prime Seafood. Each brand delivers a fully branded, mobile-first training experience with adaptive knowledge reinforcement, multiple study modes, and brand-specific guest service scenarios. One platform. Seven brands. Works anywhere.
A comprehensive menu knowledge platform for Olive Garden staff covering the full menu — Appetizers, Soups, Entrees, Desserts, Drinks, and Kids. Multiple study modes build retention and confidence, with a dedicated allergen training section covering every item. Includes guest service scenarios spanning allergies, dietary restrictions, upsells, and pairing recommendations. Works fully offline on any device.
A complete beverage knowledge platform for Olive Garden staff covering 43 drinks across wines, beers, cocktails, margaritas, and sangrias. Multiple study modes build lasting retention, with guest service scenarios that train real pairing conversations and upsell skills. Includes a smart recommendation engine that matches beverages to guest preferences and food orders. Works fully offline on any device.
Multi-strategy execution engine with automated drift governance and live performance monitoring
Mastering MCP, multi-tool orchestration, and LLM-powered development workflows at scale
Building and releasing practical tools that solve real problems for real people
A thinking space, not a content feed. The questions here tend to run deep — across markets, mathematics, physics-like structure, cognition, and computation — and they often start as conversations with AI systems before they become anything written down.
This section will carry entries from the deeper end of the intellectual work — structural questions about how systems scale, what makes a trading framework genuinely intelligent, and how patterns recur across domains that have no obvious relationship. Published when the thinking earns it.