Strategic Project Lead

Huzzle.com; San Francisco (CA); Hybrid; USD 140,000 - 210,000; Full time

Job summary

Huzzle Labs is seeking an experienced project lead to own frontier AI projects from start to finish. You will serve as the primary liaison between the lab and our expert pool, recruiting, onboarding and training domain experts, and ensuring quality across every stage of delivery.

Timelines are short and the results are highly visible. You will turn evolving client asks into actionable work, manage access and security, and report progress with precise metrics to stakeholders daily.

Qualifications

• Experience leading cross-functional projects in AI or research settings.
• Strong written and verbal communication under tight deadlines.
• Ability to map vague client asks to concrete tasks for experts.

Responsibilities

• being the lab's go-to person day to day, and keeping expectations realistic on both sides
• recruiting, onboarding and training domain experts, incl. writing the materials that get them to a consistent standard fast
• running QA: picking reviewers, keeping an eye on review throughput, spotting the mistakes that keep coming up and feeding fixes back to the pool
• reporting progress as often as the client wants (often daily). Real numbers, not ballparks
• noticing early when someone's underperforming, retraining them if that'll work, replacing them if it won't
• turning vague or shifting client asks into instructions an expert can just pick up and do
• handling platform access, permissions and offboarding, plus the security stuff that comes with them

Skills

Judgement

Communication

Ownership

Ambiguity tolerance

Detail orientation

Job description

Huzzle Labs is an applied research lab. We build reinforcement learning environments for frontier AI labs, mostly computer use, enterprise workflows and code. Basically the long, messy, multi-app tasks that make up most real professional work, and that today's models are still pretty bad at.

We've got an expert network of 300k+ vetted professionals through the Huzzle platform. We also train our own computer-use models on the environments we build (right now they're #1 in their size class on public leaderboards).

the role

You'll own projects with frontier AI labs, start to finish.

That covers a lot. You're the one person between the lab and our expert pool. You find the experts, get them up to speed, run quality, report on progress and own the relationship. Lab requirements often change mid-project. When they do, you work out what it means for the team and make sure both sides know.

Timelines are short and the results are very visible.

what you'll actually be doing

• being the lab's go-to person day to day, and keeping expectations realistic on both sides
• recruiting, onboarding and training domain experts, incl. writing the materials that get them to a consistent standard fast
• running QA: picking reviewers, keeping an eye on review throughput, spotting the mistakes that keep coming up and feeding fixes back to the pool
• reporting progress as often as the client wants (often daily). Real numbers, not ballparks
• noticing early when someone's underperforming, retraining them if that'll work, replacing them if it won't
• turning vague or shifting client asks into instructions an expert can just pick up and do
• handling platform access, permissions and offboarding, plus the security stuff that comes with them

you might be a good fit if you're

• a former founder. You've built something
• a founder's associate or chief of staff who's worked across every part of a company and is used to owning problems nobody else owns
• from ops, and you've built and run processes involving lots of people
• already a strategic project lead or program person at a human data company or an AI lab

what we care about

• Judgement. Most of this job isn't written down anywhere. You'll hit situations where the right move is unclear and waiting isn't an option. We want people who decide, act and course-correct quickly.
• Communication. The one thing we won't compromise on. You'll write to a frontier lab several times a day, brief experts who've never done this kind of work, and flag problems so someone else can act on them straight away. Writing clearly under time pressure is the core of the job.
• Ownership. It's your project. Nobody's going to check whether the roster is up to date or whether a reviewer is falling behind. That's on you.
• Being ok with ambiguity. Requirements change mid-project. Sometimes the client contradicts their own spec. That's normal, and a big part of the role is absorbing it so the confusion doesn't land on the experts.
• Detail, even when it's hectic. Access permissions, hour caps, security requirements, reporting deadlines: they all have real consequences. Most of the ways this job goes wrong are small and avoidable.

Location: San Francisco (hybrid) · Type: Full-time