Y
programming on a 80286 started it all

YVES INEICHEN

Software Engineering/HPC/Deep Learning
github.com/iff

cat <<EOF

ENGINEERING SCALABLE SYSTEMS AT THE INTERSECTION OF HIGH-PERFORMANCE COMPUTING AND MACHINE INTELLIGENCE.

Passionate about distributed systems and system-level optimization — currently building and shipping production Rust across start-up backends and open-source. The challenge of squeezing performance out of complex systems never gets old.

01_PROFESSIONAL_EXPERIENCE

09/2025 — present

FREELANCE_&_STARTUP

Freelance Rust Engineer / Co-founding Engineer

  • Co-founding engineer of a startup, implementing and extending a GPU-accelerated EDA simulation library in Rust based on latest research, with CUDA backends via bindgen/FFI and Python bindings via PyO3.
  • Building a high-performance async data ingestion pipeline in Rust using sqlx, targeting high-throughput ingestion for a startup backend.
  • Developing a train-puzzle game using Bevy ECS, exploring game architecture patterns and data-driven design in Rust.
RustBevyCUDAsqlxasync
2018 — 08/2025

PTC_INC

Principal Deep Learning Software Engineer

  • Researching and implementing state of the art computer vision architectures for classification, pose estimation, and pose refinement using TensorFlow and PyTorch. Introducing multi-GPU parallelism and quantization into production training.
  • Optimizing networks for on-device inference using LiteRT/CoreML and responsible software engineering practices.
PythonTensorFlowPyTorchDockerAWSKubernetes
2016 — 2017

SELF-EMPLOYED

Independent Engineer / Founder

  • Developing a high-throughput network analytics pipeline as a proof-of-concept for own startup, focusing on real-time stream processing and scalable ingestion at line rate.
  • Deepening and updating deep learning theory and hands-on framework knowledge to stay current with rapidly evolving research.
PythonTensorFlowNetworking
2013 — 2016

IBM_RESEARCH

Postdoc and Research Staff Member

  • Developing distributed randomized linear algebra algorithms in libskylark, scaling numerical ML workloads across large clusters using MPI and OpenMP.
  • Building graph and text analytics pipelines for large-scale data processing.
C++OpenMP/MPIPythonJavaKafkaNLP

02_EDUCATION_&_HONORS

PhD Computer Science

ETH ZURICH | 2010 — 2013

"Parallel multi-objective optimization with applications to particle accelerator design"

Master in Computer Science

ETH ZURICH | 2002 — 2008

Computational science with a minor in compiler design, focusing on parallel numerical simulations (HPC)

HONORS
  • [01]Gordon Bell Prize — "An Extreme-Scale Implicit Solver for Complex-PDEs: Highly Heterogeneous Flow in Earth's Mantle" (2015)
  • [02]PRACE Award — "A Fast and Scalable Low Dimensional Solver for Charged Particle Dynamics in Large Particle Accelerators" (2012)
SKILLS
RustPythonC++NixMPIOpenMPCUDATensorFlowPyTorchCoreML/LiteRTDockerAWSGit/JujutsuLinux/NixOSbash/zsh

03_PROJECTS

AVE.RS

Async Rust backend using axum, implementing operational transformation (OT) for collaborative climbing gym management systems through a single API.

github.com/iff/ave.rs
RustTypeScriptFirestoreNixDockerGCP

OSH-OXY

Simple fuzzy finder TUI to search shell histories, building on ratatui and the skim fuzzy matcher algorithm.

github.com/iff/osh-oxy
RustNixGit

NIHILISTIC-NVIM

My cosy neovim configuration with lots of customizations for a comfortable development environment.

github.com/iff/nihilistic-nvim
LuaNixGit

FLAKERS

Turns Nix flake lock update commits into human-readable markdown changelogs with compare links, designed to run as a step in GitHub Actions workflows.

github.com/iff/flakers
RustNix

04_BEYOND_CODE

Coffee enthusiast who co-founded a coffee roastery, combining passion for quality coffee with entrepreneurial spirit.

Ergonomics optimizer focused on split keyboards (Moonlander and Voyager), constantly refining layouts and keybindings for efficiency and comfort.

Family life, bouldering, sport-climbing, photography, and nurturing curiosity by acquiring new skills and knowledge.