01
Map
I map the system start to finish. What people do every day, where the hours actually go, what data they touch, what compliance will and will not allow. Not one task in isolation. How that task connects to everything around it.
Forward-deployed engineering for investment firms
I Build AI Systems for Investment Firms
I spent a decade investing at Citadel, Coatue, and Goldman Sachs. Then almost a decade building products, first at my own fintech, then at Perplexity AI. Now I do both at once. I embed with hedge funds, PE firms, and family offices and build the tools their teams actually use.
12+ institutional clients · $100B+ combined AUM · Systems in production, not slide decks.
NYC
Headquartered
100+ trained
Claude Code analysts
2-4 weeks
To first working system
12+ clients
Institutional investors served
Positioning
Not a Consultant. A Builder.
Most AI consultants hand you a strategy deck and leave. I ship working software.
I embed with your team and learn how you actually work. Then I build the tools that do it for you. Agents, dashboards, screens, monitors. Everything runs on your infrastructure, connects to your data vendors, and comes with documentation your team can maintain after I step back.
The difference is I have sat in the seat. I have built earnings models at 2am before a print. I have written the coverage notes. I know what matters because I did the job. Then I spent a decade learning how to automate it.
Systems-Level Thinking
How I work
Process mapping is not a new idea. V. Daniel Hunt wrote the book on it in 1996, long before any of this was AI. The discipline still holds up. You map how work really moves, find where it stalls, and fix that part. What changed is the cost of the fix.
So I do not start with the tool. I start with the system. That is true of research, and it is just as true of diligence, reporting, fundraising, onboarding, IR, and the back office. If it has steps, it can be mapped. If it can be mapped, most of it can be built.
02
Build
Working software in 2 to 4 weeks, not a proposal. I build inside your environment and test on real data with your team from day one. Ship early, fix fast.
03
Hand off
A production system your team owns. Documentation, training, handoff. Then I step back. Retainer if you want more built later.
Forward-deployed,
not staffed out.
I work alongside your team as an embedded engineer. Every engagement is hands on keyboard. I write the code, build the integrations, and ship the thing myself.
What I Build
Products, Not Presentations
01 / CO-WORKERS
Virtual Co-workers
I have launched a virtual co-worker at three firms. It lives inside Slack or WhatsApp, so there is nothing new to log into. You pass off work by collaborating with it the same way you would with a teammate, right where the work already happens.
02 / DASHBOARDS
Custom Dashboards
Portfolio analytics and position monitoring, built around how your firm actually works. A family office and a hedge fund want very different views of the same holdings. And it does not stop at the firm level. I build these down to the individual, so every PM and analyst opens the slice they care about.
03 / AUTOMATION
Workflow Automation
The repetitive work, automated start to finish. Transcript analysis, data pulls, report generation, email triage. I connect to Bloomberg, S&P, broker research, alt data, whatever your process runs on.
Work with a Builder Who’s Done Both
AI Experience
Took Perplexity Finance from Zero to One
I was General Manager of the Finance vertical at Perplexity AI. I built perplexity.ai/finance from nothing into a real competitor to Google Finance and Yahoo Finance, in under a year.
- Negotiated and bought data from every key vendor
- Worked with every model to get differentiated output
- Ran the enterprise offering inside Perplexity Finance
Entrepreneurial Experience
Self-Taught Engineer Who Founded and Sold a Fintech
I built, scaled, and sold my own fintech company. I taught myself to code to do it.
- Proficient in Python
- Have managed engineers
- Track record of building hard things
Investing Experience
Analyst at Top Global Tech Hedge Funds and a Global Bank
I was an investment analyst at Citadel, Kang Global (a Tiger Management seed), and Coatue. I started at Goldman Sachs in investment banking. Ten years on the buy side, then ten building product. That combination is the whole point.
- I have used this technology to solve real problems
- I can navigate a complicated organization
- A decade investing means no nonsense building
Clients
Current Client Roster
Anonymized by design.
Segment
Clients
AUM Range
Private Equity
2 firms
$5B - $30B
Hedge Funds
6 firms
$1B - $10B+
Venture Capital
1 firm
$10B
Family Offices
2 offices
$1B+
University
1 Ivy League institution
—
Investment Banking
1 large bank
—
13 active institutional clients across PE, HF, VC, family offices, and academia.
Case Study · My Own Research Dashboard
What a Build Looks Like: 98-Company Coverage Dashboard
This one is mine. I built it for myself, and it is a working example of what a fund could run. Agents keep it current. I do not update it by hand. It:
01
Scores 98 securities across 6 dimensions (financial trends, thematic exposure, management quality, sentiment, risk, price) with a weighted composite
02
Generates earnings previews and reviews automatically. Key metrics against consensus, guidance patterns, beat and miss history, catalysts, street Q&A.
03
Monitors material news in real time and flags what actually hits the thesis
04
Tracks an earnings calendar with linked preview and review documents for every reporting period
Built with Claude Code, Bloomberg, S&P Capital IQ, and my own scoring models. Prep on a name went from about four hours to under fifteen minutes.
This is the real thing, running live. Scroll it.
Writing
Selected writing
LINKEDIN · AUGUST 2026
Grok Is Converging With Frontier Models, Sometimes It's Even More Creative
I asked Claude, Codex, and Grok's best agentic models to analyze an earnings report: the results were hard to tell apart.
Read →EVERY · MARCH 2026
Build Your Own Financial Analyst With AI
A step-by-step guide from ChatGPT earnings previews to a custom investment dashboard, covering four progressive tiers of AI tooling for investment analysis.
Read →EVERY · FEBRUARY 2026
How Claude Code Is Transforming Finance — Without Turning You Into a Coder
How finance professionals can automate meeting prep, earnings analysis, screening, and event analysis using Claude Code and pre-built plugins, no engineering team required.
Read →EVERY
An Executive's Guide to Implementing AI
A five-step framework for moving from AI experimentation to organizational capability. Covers the "Get fluent → Assign champions → Pick one workflow → Build to 95% → Scale" loop.
Read →Questions worth asking
Do you work with firms that already have internal data science teams?
Yes, and it usually goes well. Their data scientists are focused on alpha models. I build the operational layer around them.
How do you handle compliance and data security?
Everything runs on your infrastructure. I never touch portfolio data. Every system has audit trails and a human review gate.
What's the typical engagement length?
Larger clients tend to work on a project basis, while smaller clients work best with retainer models.
Is this just prompt engineering?
No. I build production software. APIs, dashboards, agent pipelines, data integrations. The model is one piece of a much bigger system.
Contact
Let’s Build
Most engagements start with a 30 minute call. No pitch deck. Just a conversation about how you work and what is worth automating.