Roshni Kumari

How MLflow makes LLM applications observable, debuggable & measurable The hardest part of building an AI application is often not getting the model to answer. It is figuring out why it answered that way. You build an LLM application. The response looks fine. Then you change the prompt. Suddenly, one answer improves, another becomes vague, a tool call starts taking longer, token usage jumps & nobody can explain what actually changed. Traditional application logs are rarely enough to answer that question....

Open AI Accelerates Innovation. But Can It Keep Up With Autonomy? Open source made software faster. What happens when it does the same for autonomous AI? For nearly three decades, open source has been the engine behind modern computing. Linux powers the internet. Git transformed software collaboration. PyTorch accelerated deep learning research. Kubernetes redefined cloud infrastructure. Each breakthrough shared one philosophy: innovation compounds when knowledge is shared. Today, artificial intelligenc...

Everyone is celebrating AI. Few are asking who will pay the trillion-dollar bill. AI is getting cheaper. It’s a statement we’ve heard repeatedly over the past year. Smaller models are becoming more efficient. Open-source alternatives are closing the gap with proprietary systems. AI assistants are now writing code, designing presentations, analyzing data, and even handling customer support. On the surface, it feels like artificial intelligence is becoming more accessible than ever. But beneath the excitem...