Staff Software Engineer, AI

About the Job

You'll be a founding member of our AI team, working directly with the CTO to automate mortgage underwriting. Your job is to take complex, open-ended problems and figure out how to solve them—sometimes that means LLMs, sometimes it's a rules engine, sometimes it's a vendor API, and sometimes it's just good code.

Mortgage underwriting involves hundreds of discrete verification, compliance, and validation steps. You'll be tackling these sub-problems one by one—each is greenfield and requires different technical approaches. You'll define the approach, architect the solution, write a significant portion of the code yourself, and ship it to production. As the team grows, you'll determine how to split up the work and set technical direction.

This isn't a pure ML role—we're not training models from scratch. It's also not about integrating off-the-shelf APIs. It's about having strong technical judgment to pick the right tool for each problem, then building production systems that work in a regulated environment.

You'll be hands-on from day one. Week one includes taking an underwriter certification course to deeply understand the domain you're automating. You'll work in our existing mortgage engine codebase, building new automation services that integrate with the platform.

What We're Looking For

Strong systems engineering foundation

You've built large-scale production systems at high-caliber technology companies

You write code that others can maintain and build on

You understand distributed systems, APIs, data consistency, and operational excellence

You've made consequential architectural decisions and lived with the results

Recent AI / ML product experience

In the past 1-2 years, you've worked on a product that uses LLMs or ML in production

You understand when AI is the right tool and when it's not

You've dealt with the reality of prompt engineering, model evaluation, or integrating third-party AI services

You don't need to have trained models yourself, but you need to have shipped AI-powered features to users

Technical judgment across problem spaces

You can look at a problem and know whether it needs an LLM, optimization algorithm, rules engine, or just solid code

You're comfortable evaluating build vs. buy decisions and vetting third-party tools

You've worked on problems where the right solution wasn't obvious from the start

Comfort with regulated domains

Ideally you've worked in fintech, healthtech, or another regulated space

You understand that moving fast doesn't mean breaking things

You can absorb complex domain knowledge and apply it to technical decisions

Hands-on execution

You build systems yourself, not just design them—you'll be writing ~50% of the time

You can take an ambiguous goal and turn it into a working system

You're the founding member of the team, so there's no one else to delegate to initially

Basics

Job title : Staff Engineer, AI

Location : Palo Alto, CA

Stock options : own a piece of the company and we all win together

Health insurance, 401K, dental, etc.

Our technology stack :

We don't require that you've worked with any of these technologies before, this is just our stack for your information :

Languages : TypeScript (primary), with flexibility for the right tool per problem

Data : PostgreSQL

Infrastructure : AWS

About you

You :

Have strong opinions that you're willing to defend. We have a culture of vigorous discussion and debate on technical decisions. We'll push you to defend your choices, and we want you to push back.

Don't settle. Challenge yourself to frequently and consistently deliver exceptional code. If something could be better, take the initiative to improve it.

Have great ideas, and lots of them. You should see opportunities all around you to make the code, product and infrastructure better. We'll give you an environment where you can act on those ideas.

Are self-motivated. You can take a goal and drive towards it without needing extensive hand-holding. The team is supportive and loves to share knowledge and advice, but there's no time for micromanaging your work.

Are comfortable with ambiguity. There's a million ways to do things; you should feel at ease making a decision under uncertainty while balancing competing constraints.

Are confident you can learn quickly. Mortgage is complex, our platform is complex, good software engineering is complex. You've got to have an attitude that you can absorb it, get on top of it, and build something better than what came before.

Don't underestimate the domain. You'll take an underwriter course to understand what you're automating. You recognize that mortgage is genuinely complex and you need to understand it deeply to build the right systems.

About the Team

What we're not :

Building a CRUD app :

We're building a platform that's only possible due to deep technology investments and thinking hard from first principles. Cargo-culting what's come before won't be sufficient.

If you're looking to do mostly plumbing / lego piece assembling, Pylon may be very frustrating for you

A ChatGPT wrapper :

Mortgage is a sprawling mess of problems, and there isn't one tool that solves every one of them. We use LLMs, but we've reached for tools in multiple disciplines like mixed-integer linear optimization and programming language research.

An easy job :

We're building a lot of things from the ground up for the first time. Working at Pylon is like a research project where you have to ship to intelligent, opinionated customers regularly.

It's basically guaranteed you'll be handed a task that is too difficult for you to do. You might fail sometimes. You might have no idea where to start. Our team leans heavily on each other, but there's no getting around the difficulties.

What we are : A small team :

We don't have an army of engineers. If you find something is broken, you are probably the best one to fix it

All the code we write has to punch above its weight in maintainability and toil reduction

If you have a good idea, you have much more ability to put it into action than at a large company

We need generalists : specialization is for big companies that already have everything figured out. If you're smart, flexible and like getting into everything, we want you.

Working in a regulated space :

Mortgage is regulated both federally and at the state level.

We move fast, but breaking things isn't an option.

About Pylon

The $13 trillion mortgage industry at the core of the American economy runs on broken assembly lines with human-powered workflows, stitched-together software, and a series of capital markets intermediates. The costs to originate are at an all time high despite foundational shifts in foundational technology.

Pylon is rewiring mortgages from the ground up. We are building the only API-first, programmatic infrastructure that fully automates credit, compliance, capital, and operations. For the first time, originators can build and scale mortgage businesses entirely through software, not people. Our team comes from Stripe, Better, and Affim, and we are backed by Conversion Capital, QED, Citi, Fifth Wall, Peter Thiel, and the founders of Ramp, Mercury, Blend, and others.

#J-18808-Ljbffr