{"schemaVersion":"jobsearcher.job.v1","id":"ba4f4dbf9f8ea02c1bf52ef1","url":"https://jobsearcher.com/jobs/ba4f4dbf9f8ea02c1bf52ef1","canonicalUrl":"https://jobsearcher.com/jobs/ba4f4dbf9f8ea02c1bf52ef1","title":"RESEARCHER, POST-TRAINING","description":"ABOUT THE COMPANY\nWe're building autonomous research agents for recursive self-improvement (multi-agent systems that propose, run, and analyze machine learning experiments). We're a small team based in San Francisco, on-site\nABOUT THE ROLE\nYou'll lead our work on model post-training: supervised fine-tuning, preference data, reinforcement learning from human and AI feedback, reward modeling, and the evaluation suites that tell us what's actually working. You'll own a research area that meaningfully shapes our model behavior and capability.\nThis is a hands-on senior research role. You'll set direction, run experiments, and ship into production. You'll partner with the data, infrastructure, and engineering teams to make the post-training pipeline reliable and fast: improvements there compound into every model we ship.\nWHAT YOU'LL DO\nLead post-training research: SFT, RLHF/RLAIF, RLVR, DPO and successor methods, reward modeling, preference data design\nDesign and curate the data that goes into post-training (from sourcing, to filtering, to quality assessment)\nBuild and maintain the evaluation suites that measure what matters; resist Goodharting your own benchmarks\nRun rigorous experiments (controls, ablations, statistical significance) and write up internal findings clearly\nScale data pipelines and the infrastructure team to scale training\nIdentify and characterize failure modes (reward hacking, distribution drift, eval saturation) and design experiments to address them\nStay current on the post-training literature; bring useful methods in, ignore the noise\nWHAT WE'RE LOOKING FOR\nStrong track record of post-training research (SFT, RL, reward modeling) at a frontier-model lab or equivalent\n5+ years of hands-on ML research experience\nComfort with large-scale data curation and preference-data pipelines\nExperience designing evaluation suites for capabilities that aren't easily benchmarked\nFluent in PyTorch or equivalent; comfortable at the scale of distributed training\nStrong statistical instincts: you'd notice a flawed comparison before someone else points it out\nStrong written communication\nNICE TO HAVE\nPhD in ML, statistics, CS, or adjacent\nPublished research at NeurIPS, ICML, ICLR, COLM, RLC, or comparable venues\nExperience with reward hacking detection, scaling reward models, or RLHF infrastructure\nSynthetic data generation experience\nBackground in RL math (policy gradients, importance sampling, off-policy methods)\nOpen-source contributions to post-training infrastructure\nTHIS ROLE IS PROBABLY NOT FOR YOU IF\n- You're primarily interested in pretraining (that's a different role)- You'd rather invent novel methods in isolation than ship them into a model that real users run\nYou prefer benchmarks that are stable to evaluation work where the right answer isn't yet defined","company":"Makermaker","rawCompany":"makermaker","city":"Millbrae","state":"CA","isRemote":false,"isActive":false,"createdAt":"2026-07-21T11:49:19.686Z","occupations":[{"code":"15-1221.00","title":"Computer and Information Research Scientists","slug":"computer-and-information-research-scientists"},{"code":"15-2051.00","title":"Data Scientists","slug":"data-scientists"},{"code":"15-1252.00","title":"Software Developers","slug":"software-developers"}],"industries":[{"code":"541715","title":"Research and Development in the Physical, Engineering, and Life Sciences (except Nanotechnology and Biotechnology)","slug":"research-and-development-in-the-physical-engineering-and-life-sciences-except-nanotechnology-and-biotechnology"},{"code":"541511","title":"Custom Computer Programming Services","slug":"custom-computer-programming-services"},{"code":"541720","title":"Research and Development in the Social Sciences and Humanities","slug":"research-and-development-in-the-social-sciences-and-humanities"}],"jobPosting":{"@context":"https://schema.org","@type":"JobPosting","title":"RESEARCHER, POST-TRAINING","description":"ABOUT THE COMPANY\nWe're building autonomous research agents for recursive self-improvement (multi-agent systems that propose, run, and analyze machine learning experiments). We're a small team based in San Francisco, on-site\nABOUT THE ROLE\nYou'll lead our work on model post-training: supervised fine-tuning, preference data, reinforcement learning from human and AI feedback, reward modeling, and the evaluation suites that tell us what's actually working. You'll own a research area that meaningfully shapes our model behavior and capability.\nThis is a hands-on senior research role. You'll set direction, run experiments, and ship into production. You'll partner with the data, infrastructure, and engineering teams to make the post-training pipeline reliable and fast: improvements there compound into every model we ship.\nWHAT YOU'LL DO\nLead post-training research: SFT, RLHF/RLAIF, RLVR, DPO and successor methods, reward modeling, preference data design\nDesign and curate the data that goes into post-training (from sourcing, to filtering, to quality assessment)\nBuild and maintain the evaluation suites that measure what matters; resist Goodharting your own benchmarks\nRun rigorous experiments (controls, ablations, statistical significance) and write up internal findings clearly\nScale data pipelines and the infrastructure team to scale training\nIdentify and characterize failure modes (reward hacking, distribution drift, eval saturation) and design experiments to address them\nStay current on the post-training literature; bring useful methods in, ignore the noise\nWHAT WE'RE LOOKING FOR\nStrong track record of post-training research (SFT, RL, reward modeling) at a frontier-model lab or equivalent\n5+ years of hands-on ML research experience\nComfort with large-scale data curation and preference-data pipelines\nExperience designing evaluation suites for capabilities that aren't easily benchmarked\nFluent in PyTorch or equivalent; comfortable at the scale of distributed training\nStrong statistical instincts: you'd notice a flawed comparison before someone else points it out\nStrong written communication\nNICE TO HAVE\nPhD in ML, statistics, CS, or adjacent\nPublished research at NeurIPS, ICML, ICLR, COLM, RLC, or comparable venues\nExperience with reward hacking detection, scaling reward models, or RLHF infrastructure\nSynthetic data generation experience\nBackground in RL math (policy gradients, importance sampling, off-policy methods)\nOpen-source contributions to post-training infrastructure\nTHIS ROLE IS PROBABLY NOT FOR YOU IF\n- You're primarily interested in pretraining (that's a different role)- You'd rather invent novel methods in isolation than ship them into a model that real users run\nYou prefer benchmarks that are stable to evaluation work where the right answer isn't yet defined","datePosted":"2026-07-21T11:49:19.686Z","dateModified":"2026-07-21T11:49:19.686Z","hiringOrganization":{"@type":"Organization","name":"Makermaker","sameAs":"https://jobsearcher.com"},"jobLocation":{"@type":"Place","address":{"@type":"PostalAddress","addressLocality":"Millbrae","addressRegion":"CA","addressCountry":"US"}},"identifier":{"@type":"PropertyValue","name":"JobSearcher","value":"ba4f4dbf9f8ea02c1bf52ef1"},"url":"https://jobsearcher.com/jobs/ba4f4dbf9f8ea02c1bf52ef1"}}