Finance · Software · AI
Rishabb Shankar.
I build financial models and software that make the underlying logic visible.
01 / Financial analysis · −31 m
Start with the assumptions.
My finance research connects operating assumptions to valuation.
The United Airlines workbook models capacity, fuel costs and cash flow; the Apple workbook separates Services mix from segment margins. Each workbook reconstructs its analysis from filings and explicit assumptions.
02 / Software · −54 m
Build the reasoning in.
Spreadline turns synthetic financial inputs into coverage ratios, explicit policy checks and an explainable underwriting memo. It is a prototype for examining the calculation and the factors behind each result.
I care about the evidence a system leaves behind when something goes wrong.
03 / AI systems · −81 m
Eleven agents, one operator.
The marketing agency I built runs as a multi-agent system.
Nothing publishes, sends or spends on its own, and the agent that takes inbound customer messages can only triage them, because the reply should come from a person.
04 / The language model · −122 m
Every answer, accounted for.
Izumi is a conversational workspace that draws on several language models under one versioned constitution. It keeps a record of which prompt built each reply, which model answered and why.
For now it is a private project and not ready for the public, so access is by invitation.