Senior Applied AI Engineer
- Location
- San Francisco, CA
- Employment
- Full Time
- Work model
- On-Site
- Level
- Senior
- Posted
Skills
About this role
About Handshake Handshake was founded on a simple belief that everyone deserves a path to a great career, regardless of where they went to school or who they know. Today, we power 25 million job seekers, 1 million+ employers, and 1,600 educational institutions. In 2025, we started Handshake AI and built the fastest-growing AI data business in history. We work directly with frontier AI lab researchers to create evaluations, publish benchmarks, and push the boundary of data. We believe that data spend for AI training will increase by 3-5x in the next few years and continue for much longer as models take on new domains. Handshake AI supports all of the frontier AI labs, working on their most complex data at the largest scale. We’ve grown from $0 to ~$1B run rate and pay ~$60M to over 30K individuals every month.
About the Role
Frontier data for post-training models is expensive to generate and hard to measure. Labs need stronger evidence that the data they receive is accurate, useful to improve models performance, and worth the cost of producing it. We’re building an Evals platform to measure and improve the quality of expert human data and synthetic data at scale, making Frontier Model training possible and advancing AI capabilities. We are looking for exceptional Applied AI engineers who enjoy the ownership from problem definition through production delivery. You’ll work closely with researchers and customers, bring rigor to experimentation, and turn promising ideas into reliable products used at scale.
What you'll do
You will have a rare opportunity to deeply understand how frontier-grade data is generated and evaluated, and build the products and platform to do this at scale so the data we deliver to labs helps models reason more reliably, use tools effectively, and complete complex real-world tasks. Build a best-in-class evals platform. Work with operators, researchers, and labs to turn ambiguous quality needs into measurable outcomes and reusable products. Design and deploy production agents. Build agentic workflows for generating, reviewing, investigating, and improving data, with sound tool use, orchestration, human oversight, and failure recovery. Make quality measurable. Build graders, programmatic verifiers, and reward signals, calibrated against expert judgment, across LLMs, agents, multimodal models, and RL environments. Own the improvement loop. Develop eval datasets, benchmarks, and agent harnesses; analyze traces, diagnose failure modes, and run experiments until performance moves. Lead and raise the bar. Own projects from concept through rollout with product, design, operations, and research; set technical direction; mentor engineers and drive technical excellence for the team and the entire engineering org. Who we're looking for: We’re looking for engineers with 5+ years of experience who bring different perspectives, take ownership, exercise sound judgment, and hold themselves to an exceptionally high standard. A proven product builder. You have substantial experience shipping and operating high-quality software, owning decisions across the stack, and improving products based on how people actually use them. Deep expertise in agentic workflows and applied AI. You’ve built agents in production and understand tool calling, context management, orchestration, retrieval, human review, and the practical tradeoffs that determine reliability. Strong engineering fundamentals. You write excellent Python and can build scalable services, APIs, and data pipelines, with disciplined testing, debugging, and system design. An evaluation-driven approach. You know how to choose meaningful metrics, build representative test sets, assess graders, distinguish real improvements from noise, and iterate until results improve. Fluency in frontier AI. You understand modern models, agent harnesses, benchmarks, evaluation techniques, and how training data and post-training methods influence model behavior. Technical