Founding AI Engineer

a16z-speedrun; San Francisco (CA); On-site; USD 180,000 - 240,000; Full time

Job summary

a16z-speedrun is building an autonomous loan origination and processing platform for commercial real estate lenders. We seek a founding AI/ML engineer to own the intelligent systems at the heart of our product, working directly with the founders and customers and handling data, model selection, evaluation, serving, monitoring, and iteration.

You'll ship production AI systems, develop models for structured and unstructured financial data, design evaluation methods, and build ML infrastructure.

Qualifications

• You have shipped ML systems used by real users.
• Strong software engineering fundamentals beyond notebooks.
• Experience evaluating, adapting, and operating modern models in production.
• Solid judgment on model quality, latency, and cost.
• Pragmatic approach using simple solutions.
• Proficiency with AI coding agents.
• Willing to live in San Francisco and work in person.

Responsibilities

• Build production AI systems for complex, high-stakes workflows
• Develop models and pipelines for understanding unstructured and structured financial data
• Design evaluation systems that measure quality, reliability, latency, and cost
• Build feedback loops turning customer usage into better models and products
• Decide when to use foundation models, specialized models, deterministic systems, or a mix
• Ship constantly, observe real-world performance, and iterate with customers
• Establish the ML infrastructure and standards that future hires inherit

Skills

Shipped ML systems

Software engineering fundamentals

Python

PyTorch

Model serving

LLMs

Retrieval

Document intelligence

Multimodal systems

AI coding agents

Job description

The commercial real estate space is ripe for disruption. The industry has exploded in size and revenue, but still runs on fragmented CRMs, legacy loan systems, email, spreadsheets, phone calls, and document portals. Legacy players are struggling to adapt to an AI-native age, while lenders and brokers are constrained by manual work.

We’re going to change that.

We’re building an autonomous loan origination and processing platform that helps commercial real estate lenders move loans from intake through underwriting with no manual work. We’re:

• An experienced team used to shipping fast - with alumni and tech leads from Meta, LinkedIn, and Rippling.

• Backed by a16z through Speedrun.

• Working with multiple paid pilots, and moving quickly from customer conversations to production software.

The Role

We’re looking for a founding AI/ML engineer to build the intelligent systems at the heart of our product.

You’ll work directly with the founders and customers and own the full production loop: data, model selection and adaptation, evaluations, serving, monitoring, and iteration. You’ll turn messy, real-world financial information into reliable decisions and actions.

You'll:

• Build production AI systems for complex, high-stakes workflows

• Develop models and pipelines for understanding unstructured and structured financial data

• Design evaluation systems that measure quality, reliability, latency, and cost

• Build feedback loops that turn customer usage into better models and products

• Decide when to use foundation models, specialized models, deterministic systems, or a combination

• Ship constantly, observe real-world performance, and iterate with customers

• Establish the ML infrastructure and standards that future hires inherit

What We’re Looking For

• Evidence that you’ve shipped ML systems that real users touched

• Strong software engineering fundamentals beyond notebooks and prototypes

• Experience evaluating, adapting, and operating modern models in production

• Sound judgment about model quality, failure modes, latency, and cost

• Pragmatism: you use the simplest approach that reliably solves the problem

• Fluency with AI coding agents and the ability to use them without outsourcing your engineering judgment

• Live in San Francisco and are available to work in person with the team 5 days a week.

Experience with Python, PyTorch, model serving, LLMs, retrieval, document intelligence, or multimodal systems is helpful, but the ability to learn and execute matters more. You don’t need prior experience in lending or commercial real estate.