Últimas discusiones públicas del sector Datos e IA en ZIWM. Explora preguntas públicas, debates entre colegas y temas profesionales locales en industrias de EE. UU. como USPS, contabilidad, construcción, salud, transporte, e-commerce, legal, bienes raíces, restaurantes y tecnología. Inicia sesión para leer publicaciones completas y participar.
tech-data-ai
我见到不少做 BI、报表的人想往 AI 工程师转，真正难的不是会不会调模型 API，而是数据清洗、特征理解、评估指标和上线后的效果追踪。只会写 prompt 很快就到天花板。现在公司里招这类岗位，你们更看重 Python/SQL 基础，还是机器学习项目经验？
tech-data-ai
One dashboard I inherited was technically correct and still useless. Sales used one number, finance used another, and ops had a spreadsheet that everyone secretly trusted more than the BI tool. The fix started with a me…
tech-data-ai
RAG demos are easy to make look good. Production is where the weird cases show up: stale docs, two pages saying different things, an answer that sounds confident but skips the one constraint the user actually needed. Fo…
tech-data-ai
I spent a week chasing a model issue that turned out to be a data issue. Offline metrics looked decent, but production scores jumped around because one of the daily aggregates landed late on Mondays. The useful fix was …
tech-data-ai
我做数据和模型项目踩过最多的坑，不是算法不够高级，而是训练数据、线上数据、报表数据三套口径各说各话。离线看 AUC 很漂亮，上线后一查，线上特征少了一段清洗逻辑，分数直接漂。 后来我习惯先盯三件事：标签怎么来的，特征有没有线上离线一致，模型输出有没有业务能看懂的兜底。尤其是人工标注的数据，别只看数量，要抽样看错标、漏标和边界样本。很多模型问题其实是数据生产流程的问题。 上线之后也不能只看平均分。要看分桶命中率、人工复核通过率、延迟、空…