I'm new to the PM role. I'm not sure how to approach the roadmap of a ModelOps team with it being a newer function. For context, I'm at a tiny startup, we build apps that sit on top of OpenAI/HuggingFace etc models. There are no users yet, but we're cold reaching for trial users. Despite it being ModelOps team, we don't build or serve our own models, so what I know about ModelOps (data collection, model training, model validation, etc) don't really happen here. My focus should be making sure the external models like bloom, whisper, the gpt's, are supporting our app use cases (such as text summarization or generative content app), and potentially more, but I don't know what more I should focus on. Some things I've been asking myself: 1. we don't own the models (bloom etc) so even if I get feedback, I can't improve the output. 2. we're barely at a stage to scale, should I focus on growth/adaption? 3. what do you do with ModelOps if you don't serve your own models? CI/CD? Anything is helpful.. #product #productmanager #pm #modelops #lmops #generativeai #machinelearning #mlops #ml #bloom #bloomz #whisper #flant5 #gpt3 #gpt4 #gptneo #gptneox #curie #davinci #huggingface #openai #aipm #mlpm #largelanguagemodel #languagemodel
Following