AI Product Manager

 You will own the product lifecycle for AI capabilities, from discovery and requirements to launch and iteration. You will define success metrics (offline evaluation and online impact), align cross-functional teams, and operationalize data workflows including training data quality, annotation guidelines compliance, content safety labeling, and human feedback loops such as RLHF and prompt evaluation. You will collaborate with engineering and MLOps to ship reliable inference services, monitor model performance improvement, manage drift and regression, and ensure responsible AI practices.

What You Will Do

• Define AI product vision, PRDs, and roadmaps aligned to user problems and business outcomes; 
• translate product goals into model objectives, data requirements, and evaluation plans; 
• partner with ML, NLP, CV, and platform engineers on LLM training pipelines, retrieval, and inference APIs; 
• design evaluation frameworks using QA evaluation, prompt evaluation, and human-in-the-loop feedback (including RLHF) to measure quality and safety; 
• coordinate data labeling programs (taxonomy, labeling instructions, gold sets, audits) to improve training data quality; 
• run experiments including A/B testing, feature flags, and iterative rollouts; 
• manage stakeholder expectations, tradeoffs, and delivery timelines across distributed teams; 
• track product analytics, model monitoring, and error analysis to drive continuous improvement.

Required Qualifications

• Mid-Senior experience leading product initiatives with ML or AI components; 
• strong ability to write clear requirements and define measurable success metrics; 
• working knowledge of ML concepts such as classification, ranking, generative models, embeddings, and evaluation; 
• experience collaborating with engineering, data science, and design to ship production features; 
• familiarity with data workflows including data labeling, annotation QA, dataset versioning, and bias analysis; 
• ability to prioritize across roadmap, technical constraints, and operational dependencies.

Preferred Qualifications

• Experience with LLM product patterns such as RAG, tool use, prompt orchestration, and guardrails; 
• hands-on understanding of RLHF, prompt evaluation, and content safety labeling; 
• exposure to NLP, computer vision annotation, named entity recognition, or multimodal evaluation; 
• experience with experimentation platforms, feature stores, and MLOps monitoring; 
• background working with AI labs, tech startups, BPOs, or annotation vendors; experience operating in remote, global, or Brazil-based teams.