Simon Jonas Bührer
Making stuff work
Electrical engineer (ETH Zürich) working on hardware-native AI. I am trying to change the usual paradigm, where we learn in high-order floating point arithmetic and then compile that down until it ends up in hardware. Instead I take a shortcut and learn directly in the representation the hardware gives us: on an FPGA, that means learning the CLBs, the DSPs and the routing themselves. So far the results show a significantly smaller footprint (measured as the area the design would take on an ASIC), which translates into very low latency and very low energy per computation. I am excited to see where it leads.
Efficient inference, live
A live MNIST demo: a conventional dense neural network, run as an int8 GEMV (matrix-vector multiply), against a lookup-table (LUT) based network of logic gates. Both classify the same handwritten digits, but the LUT network gets there for far fewer boolean operations per sample.
Education
- 2024-2026
ETH Zürich, MSc Electrical Engineering & Information Technology
- 2020-2023
ETH Zürich, BSc Electrical Engineering & Information Technology
- 2015-2019
Kantonsschule Kreuzlingen, Gymnasiale Maturität
Experience
- 2025-2026
Everllence (former MAN Energy Solutions)
Internship to Werkstudent, Zürich
- 2022-2023
Fokusprojekt, Dübendorf
- 2021-2022
ETH Zürich
Teaching Assistant, Engineering Mechanics
- 2020
Bühler Group
Internship, London, UK
- 2019
Swiss Armed Forces
Panzerjägersoldat, Aufklärungskompanie, Thun
Languages
German (native), English (professional), French (limited), Russian (elementary)