I build AI models for a living, and I've done it at most stages of the industry: a research lab, my own startup, and now big tech. Right now, I'm an AI researcher at Amazon working on LLMs for coding. Before this, I was on the team building Pika's generative video models, where I owned much of the training data and infrastructure. I've even co-founded an AI voice-agent startup, Humanlike (YC S23), and built it as CTO from an empty repo to paying enterprise customers. Going further back, I led document-AI work at Snorkel AI and did reinforcement-learning research at Uber's AI lab, where most of my published papers (NeurIPS, ICLR, CVPR) come from. I also hold a Master's degree from Berkeley. The reason I coach is that I can be specific. I've trained the models and shipped the products myself, so I'd rather tell you what actually worked than give you advice you could've Googled. If you want to break into AI or get something of your own off the ground, that's what I'm good at helping with.