Alexander Amini

Amini completed his undergraduate, master’s and doctoral studies at MIT. His CV dates his Liquid AI role to March 2023 and his MIT research-scientist role to June 2022. He also co-founded Themis AI, which works on uncertainty-aware models. Earlier internships at IBM Research and NVIDIA involved learned control and confidence estimates for autonomous systems.[2][1]

Learn to drive in a reconstructed world

Amini coauthored VISTA 2.0, a simulator that builds virtual driving experiences from real-world sensor recordings. It renders camera, lidar and event-camera observations as a vehicle moves through the reconstructed scene, allowing a driving policy to encounter situations beyond the original recorded path. The paper evaluates learned policies in simulation and on physical vehicles. A simulator remains an approximation, so performance there does not by itself establish road safety.[4]

Teach the foundations behind applied systems

Amini organizes and lectures in MIT 6.S191, an introduction to deep learning. His CV lists successive course editions from 2018 through 2026. The course provides another entry into his work for someone who wants the foundations of neural networks before reading robotics papers.[2][1]