Nexusflow
Applied Machine Learning Engineer
Nexusflow, Palo Alto, California, United States, 94306
About Nexusflow.aiModern enterprise copilots & agents call for last-mile quality, enterprise-grade robustness and scalable operation cost, beyond the program orchestration toolings. Nexusflow tackles this challenge, enabling enterprises to own their workflow copilots & agents stack on top of powerful yet cost-effective compact LLMs. We train large language models and build last-mile quality dev tooling for copilot and agents on your workflows. Our team has built the open-source LLM NexusRaven rivaling GPT-4 in function calling with 100X smaller model size. Our team members are also behind the scene of Starling , the #1 ranked compact 7B chat model based on human evaluation in Chatbot Arena .PositionNexusflow is currently adding Applied ML Engineersto our team.Our Applied ML Engineers power our LLMs as well as Nexusflow’s methodologies for last-mile quality tooling for copilots and agents. They build the base layer of Nexusflow’s stack, contributing to tooling product and customer solutions.ResponsibilityDevelop LLMs targeted at powering copilots and agents built for enterprise workflowsDevelop toolings to attain last-mile quality and robustness for copilot & agents applications (especially under low volume of manually curated data)Building copilot & agent application solutions for high value customer verticalsWear many hats and collaborate with the whole team for product development, deployment and customer successQualificationRequiredi. Research or industrial engineering experience in at least one of the following aspects in the context of large language model or multi-modality models:Data curationPre trainingInstruction tuningCopilots & agents buildingCapability study and benchmarkingii. Excitement to contribute to both applied research and software engineering on productionizing the applied research outcomePreferredWorking experience in fast-pace teamsIn-depth experience in using or contributing to modern compute frameworks for LLMs (e.g. Deepspeed, Huggingface TGI, FSDP)Experience in turning applied research results into product components
#J-18808-Ljbffr
#J-18808-Ljbffr