Skip to content
aitrainer.work - AI Training Jobs Platform
Engineering Full-Time
Spectral Labs

Software Engineer (US)

Spectral Labs • San Francisco, CA

Company

Spectral Labs

Annual salary

$200k – $280k/yr

Location

San Francisco, CA

Listed

29d ago

Experience:
3 to 6 years
Workplace:
5 days in-office in San Francisco, CA
Equity:
Competitive equity

On-site in San Francisco, CA.

Send us your LinkedIn and CV. If your experience fits, we'll introduce you to the recruiter filling this role.

Apply for a referral →

The recruiter emails you before anything happens and may suggest other jobs that suit you better.

About this role

Spectral Labs is building foundation models for engineering physical systems.

We’re hiring an AI Engineer capable of building the product surface, the agent layer that sits on top of our models, and the glue that holds both together. We want someone who loves coding strategically and constantly seeks to optimize development speed. You live in Codex, Claude Code, or Cursor, and you’ve developed real craft at directing them: how you scope a task, what context you hand over, when you let an agent run and when you take the wheel back. You’d be frustrated somewhere that treated this as a side experiment rather than the way the work gets done. The expectation that comes with that use is real. One person here should cover ground that would conventionally take several.

This is a fully in-person role at our office in SF, with an expectation of at least 60 hours per week.

Responsibilities

  • Build and own the product surface end to end: the application, the APIs, and the backend that put our models in front of working engineers.

  • Build the agent layer on top of our models: tool use, orchestration, retrieval, and the loops that turn raw model output into something an engineer can edit and trust.

  • Design and maintain the evals for everything you ship. Non-deterministic output without measurement is a demo, not a product.

  • Handle the CAD-adjacent engineering the product requires: format conversion, geometry data wrangling, and the glue between our pipelines and the application.

  • Make the infrastructure calls that come with owning a product surface at an early company: deployment, observability, latency, and cost.

  • Work directly with research on what the model needs to do differently for the product to work, and with the product team on what users hit.

  • Push how we all build. You’ll set the internal bar for agentic coding workflows, and what you figure out should make everyone else faster too.

Qualifications

  • You code with Codex, Claude Code, or Cursor every day, and you have genuine craft at it. We want to hear about your workflow: how you break work down, what context you give, where you’ve learned not to trust the output.

  • You’ve shipped production software that real users depended on, and you can point to it.

  • Strong general engineering ability across the stack. You’ve built backends, APIs, and user-facing applications, and you don’t need a specialist to unblock you on any of them.

  • Hands-on experience building with LLMs in production. Not a demo or a weekend project, but something that met real inputs and real failure.

  • You’ve built or maintained evals, and you know why the first set everyone writes is misleading.

  • Good judgment about where AI tooling helps and where it produces confident garbage. Enthusiasm without that judgment is a liability at this speed.

  • Proficiency in Python plus at least one of TypeScript or C++, and comfort moving between them.

  • Comfort with a high-ownership environment and an ambiguous landscape. You’ll pick problems as much as you’re handed them.

  • Experience solving problems that stump others. You have to love the process of solving hard problems.

Strong Plus

  • You’ve built agentic systems that survived contact with real users, and you can talk concretely about what broke and how you fixed it.

  • You’ve built your own tooling around coding agents: harnesses, context pipelines, custom commands, evaluation loops.

  • Experience with generative models where output quality was contested and you had to define what good meant before you could ship.

  • Exposure to CAD, 3D geometry, graphics, or simulation.

  • Experience as an early engineer at a startup where you built the product surface from nothing.

  • Open-source contributions, or a tool people adopted without you promoting it.

  • Experience with inference serving, latency optimization, or running models in production.

  • A mechanical engineering or manufacturing background. You’d be one of very few people who understands both sides of what we’re building.

  • High GPA in a competitive undergrad program.

  • Objectively impressive achievements in any domain. Built something unusual? Won something hard? We want to hear about it.

Visa Sponsorship

We fully sponsor visas end-to-end. We’ve done it before, we have immigration counsel, and if we want to hire you, we don’t want paperwork to get in the way. If we make you an offer, we’ll put real resources and effort behind getting you here. The process can be slow, frustrating, and stressful, but we’ll handle the heavy lifting on our end so you can focus on what you’re great at.

Common questions

How do I apply for the Software Engineer (US) role at Spectral Labs? +

Use the Apply for a referral button on this page to send us your LinkedIn and CV. If your experience fits, we'll introduce you to the recruiter filling this role. They'll email you to check you're interested, then put you forward for this job or others that suit you better. It's free.

What does this Spectral Labs role pay? +

The listing gives $200k – $280k/yr.

Is this role remote? +

The listing gives the location as San Francisco, CA. 5 days in-office in San Francisco, CA.

Software Engineer interview prep

Questions Software Engineer interviewers often ask. They aren't taken from this listing.

How do you approach designing a system when the requirements are still likely to change?

I design around the parts of the problem I'm confident are stable, like the core data model, and keep the parts most likely to change, like specific business rules, isolated behind clear interfaces. Over-engineering flexibility everywhere just adds complexity to parts of the system that were never going to change.

What's your process for debugging an issue that only reproduces in production, not locally?

I add targeted logging around the suspected area and compare production data or configuration against my local setup, since an issue that's environment-specific usually traces back to a real difference between the two rather than a fundamentally different bug. I avoid making speculative changes without first confirming where the divergence is.

How do you decide when a piece of code needs a test versus when it's low-risk enough to skip?

I prioritize tests for logic with real business consequences if it breaks, or code that's likely to be touched again later by someone unfamiliar with it. Simple, rarely-changed glue code gets less coverage, since the cost of a thorough test there often exceeds the risk it's protecting against.

How do you evaluate a proposed technical solution's trade-offs, like performance versus maintainability?

I start from the actual constraint that matters for this specific system, whether that's genuinely performance-critical or whether maintainability is the bigger long-term cost, rather than defaulting to whichever trade-off I personally prefer. Optimizing for performance the system doesn't need just adds complexity nobody benefits from.

See all 10 Software Engineer interview questions →

Related roles

Spectral Labs

Browse all referral roles

Salaried roles at startups, AI companies and established businesses, all filled by referral. One application covers every role.

See all referral roles →