Sergiy Nesterenko - Quilter

Quick Overview

The presenter details the challenges of traditional printed circuit board (PCB) layout, which is tedious and error-prone, contrasting it with their company Quilter's AI-driven approach, culminating in the successful design of the first AI-designed motherboard and securing $40M in funding from top investors.

Key Points: PCB layout is described as tedious, error-prone, and universally loathed, requiring engineers to manually draw every trace, which took months for complex boards like a computer motherboard. The speaker, inspired by building electronics at SpaceX, encountered these difficulties while ensuring hardware survived high radiation environments. Quilter's biggest accomplishment is the 'Project Speedrun': designing the first fully capable motherboard designed by AI, which can run video calls, code, and Linux. The company has secured $40 million in funding from top investors including Benchmark, Index Ventures, Lip-Bu Tan (CEO emeritus of Cadence), and Tony Fadell (creator of iPod, iPhone, Nest). Quilter is currently growing rapidly, compiling about 100 designs every day. The company is actively hiring for ML/RL software engineering and physics simulations roles, specifically seeking C++ experts.

Context: The presentation is delivered by Sergiy Nesterenko of Quilter, focusing on the historical and current inefficiencies in the hardware design process, specifically Printed Circuit Board (PCB) layout, where engineers must manually draw complex routing, drawing an analogy to software engineers compiling code by hand. The speaker transitions from his experience at SpaceX, which highlighted the need for reliable electronics, to introduce Quilter's solution utilizing AI to automate this difficult step in hardware creation.

Detailed Analysis

Sergiy Nesterenko opens by drawing an analogy between modern software development and the hypothetical situation where software engineers still compiled code by hand, stressing that designing complex electronics like motherboards manually takes an extremely long time, citing months for a single board, especially under harsh conditions like those encountered while building SpaceX electronics. He then explains that PCB design involves three steps: Schematic (like writing code), Layout (the focus of the problem), and Assembly. The Layout phase is visually depicted as a dense, complex drawing process, which the speaker states is tedious, error-prone, and universally loathed, involving manually drawing every trace and ensuring no physical reality constraints, like component collisions or physics simulations, are violated. This complexity prevents rapid iteration. Quilter's major achievement is 'Project Speedrun,' where they designed the first fully functional motherboard using AI, capable of running Linux, video calls, and code, developed almost entirely by AI in just a couple of months. Following this success, Quilter reports strong traction, compiling about 100 designs daily across various industries including aerospace primes, major automotive companies, and hardware startups. The company has raised $40 million in total funding from prominent investors like Benchmark, Index Ventures, Lip-Bu Tan (ex-CEO of Cadence), and Tony Fadell (creator of the iPod and iPhone). The presentation concludes with a call to action, announcing they are hiring for ML/RL Software Engineering and Physics Simulations roles, particularly seeking C++ experts.