About Bonside

Capital is the bottleneck on local job creation

Small businesses create two-thirds of new American jobs and employ nearly half the private workforce — yet the cost to underwrite them locks good operators out of fair capital.

A laundromat owner with three profitable years and clean cash flow wants to open a second location. The $400k she needs is too small for banks to manually underwrite, so she's pushed to merchant cash advances at 40%+ effective rates that would kill her margin. The second location doesn't open. Six neighborhood jobs don't get created.

Multiply her by every restaurant, gym, salon, and fitness studio in the country: creditworthy B&M operators locked out of fair capital because the unit economics of evaluating them are upside down.

Both halves of the fix: agentic financing + the underwriting infrastructure beneath it

The financing business teaches us at the speed of real customers. The infrastructure — plugged into landlords, asset managers, and other capital deployers — gives us instant deal flow and lets us deploy partner capital into the businesses we underwrite. Each side compounds the other.

End state: the largest dataset of B&M creditworthiness in the world — fresh, accurate, queryable — and the ability to finance any B&M business at any time at a rate it actually deserves. Capital goes from months of negotiation to a moment's decision, and every good operator gets a fair shot at growing the business and the jobs around it.

Our right to win: distribution, data, and the moment of need

We've closed a $1.2T asset manager on the infrastructure side, we're live across 75+ properties, and we're pursuing embedded integrations into the moments operators need capital most — commercial real estate buildouts, tenant evaluation — where the pain is highest and the documents are already in motion. That's how we beat the unit economics math that's kept B&M financing stuck for decades. Underneath sits a proprietary B&M data layer: hundreds of thousands of normalized data points and specialized agents for extraction, normalization, risk evaluation, and pricing.

Staff AI engineer

You'll own an agentic product end-to-end, ship it, and watch real businesses get funded–-and create new jobs–-because of what you built.

What you'll do

• Build production-grade agentic systems — planning, tool use, memory, orchestration

• Take prototypes to production with real attention to latency, cost, and reliability

• Design eval frameworks that keep agent behavior trustworthy as we scale

• Work directly with founders to translate strategy into shipped product

• Find places where agentic workflows can replace manual processes — and ship them

What we're looking for

• Track record of building and commercially shipping 0-to-1 production systems, not just prototypes.

• End-to-end problem ownership; comfortable starting without a spec

• Genuine fascination with what LLMs can and can't do

• Strong Python and TypeScript

• Clear communicator on tradeoffs, risks, and decisions

Nice to have

• Hands-on experience building production LLM systems, including orchestration, tools, memory, evals, and AI/ML evaluation frameworks

• Background in fintech, underwriting, or operational systems