Machine Learning Research Engineer

We’re partnering with a frontier AI startup We’re working with an early-stage AI company that’s redefining how models learn to understand subjective quality , from writing tone and design aesthetics to emotional resonance and creative expression. They’re collaborating with leading AI labs and building new methods that help models reason about creativity, taste, and quality. The founding team is small, highly technical, and deeply curious—building at the intersection of research, product, and creativity . The Role As a Machine Learning Research Engineer , you’ll own end-to-end research cycles—designing, running, and analyzing post‑training experiments that help models evaluate style and subjective quality. You’ll collaborate directly with AI labs and creative experts, and have the opportunity to publish your work. What You’ll Do Train reward models, classifiers, and verifiers across creative domains (writing, design, visual style) Develop benchmarks and evaluation methods for subjective tasks Run post‑training (RLHF/DPO) experiments on open‑source and proprietary models Collaborate with AI labs and creative experts on pilots and research projects Publish findings through blogs and whitepapers What We’re Looking For 2–6+ years in post‑training, ML research, or ML engineering Strong experience with Python, PyTorch, LLMs , and multimodal models Proven ability to run fast, iterative experiments from data to evaluation Curiosity for subjective AI —creativity, tone, and aesthetics Low‑ego, collaborative mindset; thrives in a small, in‑person environment Bonus Experience working with post‑training data or data vendors (Scale, Labelbox, Snorkel, etc.) Personal interest in creative fields (design, art, writing, or similar) Why Join Build at the edge of AI and creativity Collaborate with frontier research labs Work closely with an inspiring founding team in Jackson Square, San Francisco Competitive salary and meaningful early‑stage equity Compensation: $200K–$350K + 0.25–1.5% equity Interview Process: We’re proud to partner with a founder who values creativity, collaboration, and curiosity above ego. #J-18808-Ljbffr