Applied AI for Finance

AI that speaks the language of finance.

I help investment firms, fintechs, and finance teams turn AI from hype into working tools — agents, research automation, and quant-grade modeling. Built and explained by someone who speaks both quant and code.

The gap

Most AI consultants don't speak finance. Most finance people don't build AI. Quant AI Lab is the intersection: a quant background, hands-on AI engineering, and the ability to explain both clearly enough to make a decision on.

Vendor-agnostic by design — I build on whichever model is best today, so every model release makes your tools better, not obsolete.

Services

Three ways to work together — from figuring out where AI pays off, to shipping the software that delivers it.

01 · ADVISORY

AI Strategy & Advisory

Cut through the hype and find where AI actually moves the needle for your firm — and where it doesn't.

  • AI Opportunity Audit (1–2 weeks, fixed scope)
  • Fractional AI Advisor (monthly retainer)
  • Model-agnostic tooling & vendor strategy
02 · BUILD

Applied AI & Agents

Working software, not slideware. Agents and AI tools built into your real workflows.

  • AI agents & workflow automation
  • RAG over your documents & data
  • Prototype sprints — a working PoC in 2–4 weeks
03 · QUANT EDGE

Quant & Data Intelligence

The differentiator a generic AI shop can't offer — modeling rigor from a finance-native builder.

  • Regime-aware risk & scenario modeling
  • Research & analysis automation
  • Data pipelines & evaluation for reliable AI
Approach

Three principles behind every engagement.

01

Finance-native

I speak quant. Risk, sequence-of-returns, market regimes — you don't have to translate your problem into plain English first.

02

Model-agnostic

Built to swap models freely. You ride the improvement curve instead of being locked to one vendor's roadmap.

03

Honest & hands-on

Real experiments with the caveats stated plainly. When something's a proxy or a limitation, you'll hear it — on the record.

Proof, not promises

Every claim I make, I show — and open-source. One example:

An unsupervised model, given only S&P 500 prices, found the market's volatility regimes on its own — and the real signal wasn't how high volatility was, but how fast it was rising.

It flagged the 2011, 2018, 2020 and 2022 stress periods with no labels and no hindsight. The caveat, stated openly: the regime fit uses whole-history data, so it carries look-ahead bias — fixing that with an expanding-window signal is the next experiment. That honesty is the point.

4,159
Trading days
3
Regimes found
0
Labels needed

Full code & write-up, open on GitHub → quant-ai-vol-regimes

How it works

A low-risk path from first conversation to shipped work.

Step 1

Start a conversation

Tell me what you're trying to do with AI — or what you're unsure about. No pitch deck required.

Step 2

Opportunity Audit

A fixed-scope, fixed-price engagement that maps the highest-ROI AI use-cases for your firm, with a clear roadmap and honest risk assessment.

Step 3

Build

Turn the roadmap into working software — a prototype sprint or an ongoing advisory retainer, whichever fits.

Step 4

Handover & iterate

You get something that works, documented so your team can run it — and a partner for the next step when you're ready.

Have an AI problem worth solving?

The first conversation is free — and useful whether or not we work together.

Get in touch