Johanna Angulo ^{Top-Executive}

I lead AI teams at the intersection of research, product, and business — translating frontier capability into systems that ship, scale, and stay defensible. In parallel, I'm a PhD researcher working on contamination-controlled safety benchmarks for agentic systems and explainability (XAI) — with four 2025 peer-reviewed publications and two published open benchmark datasets (n = 1,047) for rare-disease gene prioritization. At Multiverse Computing, I built the AI evaluation function from zero: a 16-person multicultural team owning evaluation and safety strategy across LLMs and quantum-compressed architectures. I gated 8 model releases through safety benchmarking and red-teaming, and led large-scale evaluations on the Leonardo and MareNostrum (EuroHPC) supercomputers. I also served as primary contact with independent evaluation bodies like Artificial Analysis, coordinating benchmark submissions and public results. Earlier, I led AI engineering at YeY Digital across ML, NLP, and generative AI — shipping production agentic systems with measurable ROI, including a compliance assistant that cut review hours by 55% and a multi-agent content pipeline that cut API costs by 65% through smart model routing. Before that, I directed engineering at a global multilingual NLP company serving life sciences, healthcare, and pharma, leading a 35+ person team and representing the company to clients across three continents. What I bring to senior AI roles: ● Strategic vision. AI roadmaps that align technical capability with business goals and regulatory reality. ● AI safety & evaluation. Release gating, safety benchmarking, red teaming, and human-in-the-loop evaluation.● Governance that ships. EU AI Act and ISO/IEC 42001 turned into technical controls, policy-as-code, and audit-ready evidence.● Team building. Hiring, growing, and leading multidisciplinary teams across cultures and time zones.● Stakeholder fluency. Comfortable in the boardroom, the customer call, and the model-improvement standup — translating fluently between them.● Technical depth. LLMs, agentic AI, multimodal systems, and evaluation frameworks — backed by peer-reviewed research.I am most useful where research meets product meets business — AI safety, AI evaluation, responsible AI, and AI governance leadership. If you're building teams that bridge those worlds, let's connect.