Summary
Michael Lu is a machine learning engineer in the San Francisco Bay Area with eight years of experience building custom transformer models and production ML systems that optimize CTR and CVR for paid ads. He currently drives commerce media optimization at Moloco, applying long-horizon scoring and recommendation techniques to materially increase ad spend efficiency. Michael’s background blends hands-on product work and deep research—recent projects span top-down attention steering in vision, RL-finetuned LLMs for game-theoretic reasoning, and search-augmented agents for sequential decision-making. He has also founded a vertical-AI transformation startup focused on embedding agents into business workflows, underscoring his product-first approach to automation. Comfortable moving models from lab to scale, he pairs experimentation in multimodal retrieval and opponent modeling with pragmatic optimization for ad performance. Colleagues would describe him as a researcher-engineer who targets high-impact wins that bridge academic insight and production ROI.
8 years of coding experience