Member of Technical Staff - Research

Valkai•San Francisco, CA

About The Position

You'll turn difficult problems in real workflows into research questions, develop improvements across models and agent systems, and carry results into production. This role is for a researcher with strong engineering ability who can move between model training, experimentation, and applied AI systems, choosing the right approach for our users.

Requirements

Deep ML foundations: You understand learning, optimization, probability, and statistics, and use that understanding to reason about model behavior and experimental results.
Hands-on model experience: You have trained, adapted, or improved models and can explain the decisions behind your work. You bring depth in areas such as post-training, reinforcement learning, etc.
Research judgment: You turn ambiguous problems into clear hypotheses, design experiments, and can develop new approaches.
Strong applied judgment: You stay close to real workflows and know when to improve the model, the data, or the system around it. You weigh quality, reliability, latency, and cost.
High agency and ownership: You help define the research direction, take responsibility for outcomes, and move from investigation to execution without waiting for a detailed plan.
Clear collaborators: You communicate findings and uncertainty clearly, seek context from domain experts, and help research, engineering, and product make decisions.

Nice To Haves

You have carried a research idea into a product or system used in real workflows.
You have research publications, open-source contributions, or substantial industry work in ML, LLMs, agents, or evaluation.
You have experience with reward modeling, synthetic data, or learning from human feedback.
You have worked with distributed training, inference optimization, or large-scale experimentation.
You have worked in a high-growth early-stage company.

Responsibilities

Own applied research projects from problem definition and experiment design through implementation and production validation.
Train and adapt models for the reasoning, retrieval, and decision-making tasks
Develop and test improvements to agent systems, including tool use, context, memory, planning, and recovery from failures.
Build datasets, evaluation environments, and grading methods that capture the difficulty of real workflows.
Inspect model outputs and agent traces, identify recurring failure modes, and turn those findings into changes to training, data, or system design.
Build the experimentation tools and pipelines needed to run, compare, and reproduce research efficiently.
Partner with AI Strategy, product, and engineering to understand workflows and ship improvements to production.