I build research tooling for funds. Your names, your theses, and what the public record keeps publishing about them — filings, transcripts, dockets, news — in one system that stays current on its own and tells you what changed.
The research day, as it actually runs.
The inbox at 6am is a triage exercise: what's going to kill me today? Sell-side notes, filing alerts, transcript services, news — none of it knows what you own or why you own it.
Every position went on with a thesis — a few drivers the whole case rests on. It lives in a memo and in your head. Each new filing and call either confirms it or quietly erodes it, and nothing in the stack does the diff.
Someone at the fund — you, an analyst, an intern one summer — built the dashboard, the scraper, the watchlist tool. It worked for a quarter. Then a feed changed, the person who understood it moved on, and it stopped being true.
A workbench that watches the record, with a module built around your method inside it.
The chassis. A watchlist that actually works for you: filings, transcripts, dockets, registries, and news ingested as they land, filtered against what you own and what your case rests on. It holds state — so it can answer "what changed since yesterday," not just "here's everything again." Flags come to you, and every flag links to the source line it came from.
The part built around how you invest. A claim-vs-record monitor for a forensic short book. Catalyst clocks with runway and lockup dates attached. Covenant extraction across a credit book. A management promise-vs-delivery ledger. Language baselining across the last eight quarters of calls. Your method, running continuously — not a vendor screen with your logo on it.
Where engagements deepen. The checklist that lives in your head becomes software the whole team runs — the same discipline on every name, and an answer for allocators who ask how the process survives contact with a bad quarter.
I also fix fund plumbing — reconciliation, reporting, the data spine research tooling sits on. This side starts differently: a scoped diagnostic first, then a build. If ops is your pressing problem, start there.
Most engineers don't understand funds. Most fund people can't build.
The hard part of a research monitor isn't the pipes — it's judgment. A flag is only worth your attention if it's tuned to what your case actually rests on, not keyword-matched to your tickers. That tuning is the product. I build the system around your drivers and keep calibrating it against what you confirm, dismiss, and correct.
Nothing surfaces without the source line it came from — the filing paragraph, the transcript passage, the docket entry. You can always see why the system said what it said. No black boxes; if you can't check it, you won't trust it, and you'd be right not to.
Every build lives in your infrastructure, lands with a runbook and full handoff documentation, and can be walked away from on thirty days' notice. That's also the story your COO can tell an allocator in due diligence — scoped access, no resident vendor, nothing leaves the building.
Fifteen-plus years of production software, and a working systematic trading infrastructure of my own behind the practice. I read the filings too. No three-month ramp to understand what your book is doing, and no guessing about which problems are worth solving in software.
Proof first. Then a fixed-scope build. Then it stays current.
You see it running before you spend anything.
When there's a real fit, I build a thin version on two or three of your names and demo it on the first call. Not a deck about what the system would do — the system, doing it. If it isn't useful, we've both spent thirty minutes finding out.
A working system, shipped.
The workbench core plus your module, on your book. Fixed price, fixed scope, fixed timeline — demo by week two, shipped by week six. The pilot band is for smaller funds and first engagements: same system, tighter scope. Larger shops with a real procurement process get scoped individually.
Software that watches the record needs an owner.
Feeds change, filings formats drift, your book turns over, the reads need recalibrating. Continuity covers that: the system keeps working and keeps getting sharper. Every build also lands with full handoff — your cloud, runbook, documentation — so you can run without me at any point. Most funds keep continuity; none are locked into it.
For heavier ongoing work — multiple systems, integration projects, a standing engineering lane — there's an hourly retainer at $300–500/hr, 5–10 hours a week. A finance-literate engineer at market costs $200–350K loaded; this is that capability in proportion to the work.
Fund plumbing — reconciliation, reporting, the data spine — starts differently: an operations diagnostic from $3,500, one to two weeks inside the workflow, producing a specific list of what to build, what to automate, and what to leave alone. Then a fixed-scope build if one is warranted. No obligation to continue.
Fifteen-minute call. If there's a name worth testing, I'll build the thin slice before we talk numbers. If there isn't a fit, I'll say so.
Book an intro callOr email [email protected] with a sentence about your fund and a name or two you'd want watched.
The stuff you're probably wondering.