Autopilot: Automating Circuit Board Design Using Reinforcement Learning

In this episode from Autopilot with Will Summerlin, Sergiy Nesterenko, founder of Quilter (backed by Benchmark), discusses designing PCB circuitboards end-to-end using reinforcement learning.

"Autopilot" host Will Summerlin and Nesterenko cover how current PCB boards are designed and how Quilter’s tech stack enables faster board design, what better circuit boards enable in the future, their GTM and where they are seeing most market pull right now, and much more.

TIMESTAMPS:
(00:00) Intro & Sergiy’s Background
(04:31) What Sergiy learned from SpaceX
(05:53) Founding thesis of Quilter and Quilter’s journey
(06:57) Where would one find circuit boards?
(08:11) What is the process of designing a circuit board?
(09:39) Design process today with Quilter
(14:34) Sponsor: Omneky
(16:01) Quilter’s thesis and designing more complex circuits
(18:25) How much are humans currently paid for board design
(19:41) Labour dynamics in board design
(20:30) Do most companies have board designers in house?
(21:14) Incentive structure
(22:44) What does a high performance circuit board look like vs low performance?
(26:37) Quilter’s technology stack
(29:15) Sponsor: Plumb | Squad
(31:15) How Quilter can grow with scale?
(33:50) Where is circuit manufacturing happening
(37:17) What other parts of knowledge work can be solved with reinforcement learning
(41:03) GTM and who Quilter is selling to
(42:48) Pricing
(44:04) Where Quilter is seeing the most market pull right now
(45:21) What makes Quilter an exciting company to work at or invest
(47:14) The effects of closed research in private companies for the industry
(49:25) Open source vs closed source
(51:14) What Sergiy would advise to himself in his early founder
(53:59) What drew Sergiy to working with Benchmark
(56:19) Wrap