Building AI that
aerospace can rely on

Mark Chavira, PiLogic's co-founder and CTO, spends twelve and a half years building real-time avionics operating systems, where speed, predictability, and reliability are the whole job. Alongside it, a Stanford MS and a UCLA PhD in probabilistic inference.

Mark Chavira first-authors the paper in Artificial Intelligence 172, and it becomes the standard reference for the technique this engine is built on. Exact reasoning had stalled by then: past a certain complexity the algorithms ran out of time and memory, and most of the field moved to approximation instead.

Mark Chavira co-authors NASA-funded research applying the approach to ADAPT, the electrical-power-system testbed at NASA Ames. Sub-millisecond diagnosis on models of more than 400 nodes. ADAPT is a ground testbed, not a flight system.

Mark Chavira spends thirteen years at Google as an engineering director, leading a 125-person AI engineering team building probabilistic inference systems.

Mark comes out of retirement to found PiLogic with Johannes Waldstein and Geoff Bough. The two problems that had blocked the field for decades — building models fast enough, and reasoning at the edge — were finally solvable.

Scout Ventures, Seraphim Space, Sovereign's Capital, Flex Capital, and FN Fund, with angel investor Gokul Rajaram.

A Cooperative Research and Development Agreement with the Air Force Research Laboratory, with Space Systems Command support. Testbed access at Kirtland Air Force Base, demonstrations to military customers, and TS/SCI clearance sponsorship.

PiLogic selected for the SpaceWERX AIS Challenge.

NASA selects PiLogic to test Exact AI for detecting and diagnosing failures across critical onboard spacecraft systems.