Want to know if this job is worth applying to?
San Francisco, California, United States
Onsite
Not Available
We have 20 engineers and more than $100M in ARR, and we're scaling the engineering team roughly 3x over the next year. The AI Platform team is where our data advantage becomes product: it owns the matching, ranking, and agentic systems that decide which recruiter sees which role, which candidate reaches which company, and how much of the hiring process runs without a human touching it.
We're hiring a technical leader to own the AI Platform engineering team. This is a TLM role: you'll manage a pod of Applied AI, ML, and Product engineers while staying deep in the core ML work yourself, from data and model design to the hardest ranking and matching problems. You'll partner closely with product and data science and be accountable for outcomes, not output: match quality, automation rate, and the hiring velocity our systems create for customers. Title (Tech Lead Manager, Technical Director, or Engineering Manager) depends on your experience; the ownership doesn't.
Expect to spend at least a third of your time building, more in the early days. You'll read PRs, review evals, and ship production code yourself, every week, not occasionally. The management load grows as the team does, but the technical bar you set personally is a core part of the job.
The team. Hire, coach, and level a pod of applied AI, ML, and Product engineers. We run small autonomous teams with high judgment expectations, and we promote on judgment, not lines of code.
Core ML, hands-on. The matching, ranking, and retrieval models are yours at the code level: data and labeling strategy, feature and embedding design, model selection, training, and fine-tuning. When match quality plateaus, you're the one who finds the next win.
The hardest modeling problems. The ambiguous, high-risk ML work that isn't ready to delegate: a ranking model that stops improving, an offline eval that disagrees with production, a cold-start problem with no clean labels. You take these on and ship the fix yourself.
Matching and ranking. The models at the core of the marketplace: retrieval, ranking, and personalization systems that decide how supply meets demand. Improvements here move revenue directly.
Agentic systems. LLM-powered workflows and automation that take real work off recruiters, hiring managers, and our own ops team. Reliability matters as much as capability; these systems act on behalf of real businesses.
Trust & safety. The systems that keep the marketplace honest: fraud and spam detection, submission quality enforcement, and guardrails that keep AI outputs accurate and fair when they touch real candidates and real hiring decisions.
Evaluation rigor. Nothing ships on vibes. You'll own the eval frameworks that measure quality, reliability, and business impact in production, and build the team's muscle for knowing when a model change is actually better.
The quality/cost/latency frontier. Deciding when an LLM is the right tool and when traditional ML wins, and keeping the whole system fast and economical as usage scales.
Roadmap and delivery. Turn ambiguous business goals into a quarterly plan the team believes in, then ship it.
Staff-level IC depth in ML or applied AI: you're the person others pull in when the problem is genuinely hard, and you've stayed technical while leading
1+ years managing or formally tech-leading engineers (Tech Lead Manager) or 4+ years including setting technical direction across multiple teams (Technical Director)
You've shipped production ML or LLM systems used by real people at scale: retrieval, ranking, recommendation, agentic workflows, or similar
Eval-driven: you know how to measure model quality against business outcomes, not just offline benchmarks, and you can tell a real win from noise
Data-fluent: you can pull your own SQL, sanity-check an experiment readout, and push back on a bad metric
AI fluency is a floor here, not a differentiator. You use AI tools in your own work and you know how to raise a team's leverage with them
You write things down. Operating norms, decision docs, postmortems. Clarity in writing is how small teams stay fast as they grow
You still want to build. If writing production code every week sounds like a step backward, this isn't the role
AI Platform is where Paraform's moat gets built. We have proprietary hiring data no one else has, and the systems your team builds turn it into better matches, faster fills, and automation the rest of the industry can't replicate. You'll set the technical direction for AI at the company while the org is still small enough that your decisions compound. For more info on where we're headed, and why this role is so important, check out our company vision.
Paraform is the first end-to-end AI recruiting marketplace, modernizing one of the largest and most fragmented markets in the world: hiring. We connect companies with thousands of specialized recruiters and AI agents that work together to fill roles faster and more accurately.
We believe recruiting works best when human judgment is paired with AI that truly understands hiring. That's why we treat recruiters like entrepreneurs, not vendors, and build the financial, operational, and technical infrastructure that powers their success.
Industry leaders like Decagon, Palantir, Rippling, Abridge, and Hightouch hire through Paraform, and we grew revenue 8× last year. We're backed by Scale, Felicis, A*, BOND, Liquid 2, DST Global, and the founders of YouTube, Instacart, Canva, and more.
Our mission is to make hiring fast, reliable, and scalable for every company in the world. For more on where we're headed, read our company vision.
