Systems & Research Engineer

“I’ve seen things you people wouldn’t believe.”

— Blade Runner

Now we’d like to see what you’ve built.

We’re recruiting a Systems & Research Engineer for a fast-growing US applied AI company building production systems for freight and global supply chains.

Founded by engineers from MIT and Stanford, the company raised $6M in seed funding in early2026 and is already operating AI products at serious real-world scale.

One of its core fraud and identity platforms now screens approximately 3,000 drivers every day.

The team is small.

The ambition isn’t.

And the engineering bar is deliberately high.

What You’ll Actually Do

This is not another generic AI Engineer role.

You’ll sit at the intersection of:

AI systems.

Research.

Inference.

Model serving.

Performance engineering.

Distributed systems.

Evaluation.

Your job is to understand how production AI systems actually behave.

Where is the bottleneck?

GPU?

CPU?

Memory?

Network?

Serving architecture?

Concurrency?

Model choice?

You’ll form hypotheses, build benchmarks, test alternatives and use the results to make real engineering decisions.

A recent example involved benchmarking speech-to-text approaches and building a hybrid open-source and production system that outperformed vendor alternatives on both quality and economics.

That’s the level of problem we’re talking about.

We Want the Experiment, Not Just the Percentage

A resume saying:

“Reduced inference latency by 37%.”

isn’t enough.

We want to know:

What was the baseline?

What did you think was happening?

How did you test it?

What alternatives did you benchmark?

What did the data show?

And most importantly:

What engineering decision changed because of the experiment?

You need to be able to walk us through at least one serious benchmark or experiment you personally designed and ran involving areas such as:

Model serving

Inference

Evaluation

Retrieval

Speech systems

Agent infrastructure

The strongest candidates think like researchers but ship like engineers.

The role specifically requires performance engineering on AI/model systems rather than distributed systems with no model in the loop.

The Kind of Work You Could Be Doing

Profiling production AI systems and identifying GPU, CPU, memory or network bottlenecks.

Benchmarking serving frameworks such as vLLM, SGLang and alternative architectures.

Investigating inference throughput and latency.

Evaluating open-source versus closed-source models.

Optimizing workloads across cost, quality and concurrency.

Building evaluation infrastructure.

Understanding model behavior under real production traffic.

Reasoning about distributed systems where there is genuinely a model in the loop.

Turning research findings into production architecture decisions.

And occasionally proving that everyone’s first assumption was wrong.

Who Could Be Right?

Your current title might be:

Research Engineer

ML Systems Engineer

AI Infrastructure Engineer

Inference Engineer

Performance Engineer

ML Platform Engineer

Systems Engineer

Experience inside sophisticated ML infrastructure environments is highly relevant.

Think engineering problems similar to those encountered at Google DeepMind, Meta, Stripe, Amazon AGI, Anthropic, Together AI, Fireworks AI, Baseten or similarly strong AI organizations.

But pedigree alone won’t get you through.

We’re looking for evidence of exceptional technical work.

The strongest candidates can explain their experience like this:

Here was the hypothesis.

Here was the baseline.

Here was the benchmark.

Here’s what we discovered.

Here’s what we changed.

What This Role Is NOT

This is not DevOps.

It isn’t frontend.

It isn’t conventional full-stack development.

It isn’t Web3.

And pure distributed-systems experience, however impressive, isn’t enough if there has been no meaningful AI or model component.

There needs to be a model in the loop.

AI Coding Agents

This team uses coding agents heavily.

Every day.

The philosophy is simple: excellent engineers should increasingly spend their time deciding what should be built, how it should work and whether the result is correct, rather than manually producing every line.

Your ability to work effectively with modern coding agents will be assessed during the interview process.

Remote Really Means Remote

This role is 100% remote across the USA.

New York.

San Francisco.

Austin.

Seattle.

Boston.

Miami.

Denver.

Wherever you do your best work.

The company operates asynchronously, with no fixed working-hour or timezone-overlap requirement. There is a San Francisco office available if you want it, but attendance is not required.

Compensation

USD $220K–$300K Total Compensation + Equity

USD $300K is the ceiling.

This is high-bar, opportunistic hiring rather than a volume recruitment campaign. The company is building an ongoing pipeline and wants exceptional engineers, not simply more engineers.

The Bar

One of the founders has a very simple hiring philosophy:

He wants engineers joining the company who make the existing engineering team better.

That means the bar is high.

Deliberately.

If you’ve done genuinely exceptional work around AI systems, inference, model serving, evaluation or performance engineering, we want to hear from you.

And when you apply, don’t just tell us what you built.

Tell us about the experiment.

Systems & Research Engineer | Applied AI | USD $220K–$300K + Equity

100% Remote Across the USA | Fully Async | San Francisco Office Optional

Model Serving | Inference | ML Systems | Performance Engineering