Hidden costs of hallucination: Legal and ethical risks that companies must understandA few months ago, I sat with the leadership team of a global bank that had just rolled out an internal generative AI assistant. It was…4d ago4d ago
Designing multi-agent systems for complex long-horizon tasksIf only it were as simple as plugging in an AI agent framework and watching the system come to life. I learned this the hard way last year…5d ago5d ago
Zero-shot translation: The promise and pitfalls of cross-lingual LLM generalizationAs VP of Technical Content at Educative, I often see learners and engineers eager to unlock multilingual capacity without the burden of…6d ago6d ago
Why I believe future ML engineers will spend more time on data than modelsOver the years, I’ve witnessed some major shifts in software engineering. I remember observing teams as they transitioned from monolithic…Oct 29Oct 29
Why multimodal AI is harder to productionize than you thinkWhile lecturing at universities, co-founding startups, and contributing to projects at Microsoft, I’ve witnessed many technologies struggle…Oct 28A response icon1Oct 28A response icon1
Lessons learned from treating models like codeI’ve watched brilliant data science teams build models that could transform their businesses for years, only to see them gather dust in…Oct 27Oct 27
Building knowledge-enhanced chatbots with open-source tools (a RAG tutorial)I’ve spent years navigating the evolution of artificial intelligence. I’ve seen firsthand how large language models (LLMs) have transformed…Oct 22Oct 22
Detecting fraud and connections in complex networksEarly in my career, I remember coming across fraud detection systems that relied on simple rules, such as flagging a transaction if it…Oct 21Oct 21
Unpacking the ‘data gravity’ problemWhen I think back on the machine learning projects I’ve led, whether in academia or production teams, the same story pattern often emerges…Oct 20Oct 20
Fact or fiction? My attempts at hallucination detection in LLMsHave you ever seen an AI confidently tell you something that turns out to be completely wrong? I have, and it’s both fascinating and…Oct 17Oct 17
