Curriculum Vitae · Glück auf

Born on coal, building in silicon

Senior engineering leader in long-read DNA sequencing and real-time signal processing. From the coal pits of the Ruhr to the labs of Silicon Valley. I build and scale the teams that turn petabytes of photon counts into HiFi reads that improve human health. Manager by day, programmer by night.

  • Open to select opportunities
  • Gelsenkirchen → San Francisco Bay
  • Fully remote · based in Germany

About · −340 m

The engineer behind the work.

Dr. sc. ETH Zürich in computational biology, specialized in statistical modeling of long-read sequencing. A decade at Pacific Biosciences took me from hardware-near modern C++ on a single instrument to leading the organization that owns the entire on-instrument analysis stack.

I architected the real-time data path that turns petabytes of photon counts into billions of HiFi reads, and helped launch five sequencing platforms – Sequel through Revio and Vega. I grew the instrument-analysis team from one contractor to fifteen engineers, all fully remote.

The work spans signal processing, bioinformatics, GPU/CPU optimization, and the productization of deep learning – real algorithms, shipped reproducibly, with an eye on compute cost. I'm a trusted partner to NVIDIA and Google.

I was the first at the company to bring AI into engineering, and I'm still its most demanding user. I build novel agents that cut instrument troubleshooting from weeks to minutes, and I run self-hosted open-source LLMs on repurposed hardware – putting AI in the hands of the whole software organization without the per-token bill.

5Sequencing platforms shipped
10+Years leading remote teams
864TBPhoton counts processed per day/GPU
95%Faster troubleshooting through AI
30%Revio compute cost reduction

Experience · −620 m

A decade building platforms.

Mar 2024–nowRemote · Menlo Park, USA

Senior Director, Instrument Analysis

Pacific Biosciences

Lead the on-instrument analysis organization: up to 15 direct reports architecting the full path from photon counts to highly accurate long reads. Primary technical contact for the NVIDIA and Google collaborations, with a direct line to executive management. Own hardware selection and architecture for on-instrument compute. Implement internal agents to reduce troubleshooting time from weeks to minutes. Deploy LLMs internally for the whole software organization. Daily AI user to ship products.

Org leadershipArchitecturePartnerships
Feb 2023–Feb 2024Remote

Director, Platform Bioinformatics

Pacific Biosciences

Owned all platform bioinformatics software, built 100% reproducible and deployed on and off instrument. Scaled pipelines to process hundreds of millions to a billion HiFi reads in a single run, and shaped company objectives through external partnerships.

Team leadershipReproducible buildsScale
Jul 2021–Jan 2023Remote

Principal Engineer & Associate Director, Bioinformatics

Pacific Biosciences

Led the on-instrument bioinformatics team while continuing to ship as an individual contributor, generating long, accurate HiFi reads in near real time. Ported and invented GPU algorithms in CUDA; drove architecture and hardware decisions.

Player-coachCUDA / GPUReal-time
Sep 2020–Jun 2021Remote

Principal Engineer, Bioinformatics

Pacific Biosciences

Prototyped the compute architecture for upcoming sequencing products, evaluating and building on emerging platforms including ARM, RISC-V, and GPGPU. Enhanced the customer-facing toolchain (CCS, Lima, Iso-Seq) while mentoring the team.

Emerging hardwareMentorshipToolchain
Aug 2018–Aug 2020Remote

Senior Staff Engineer, Bioinformatics

Pacific Biosciences

Technical leader across cross-functional teams. Made Iso-Seq up to 100× faster while improving accuracy, delivered structural-variant calling at the highest recall and precision, and drove down time-to-result for the CCS algorithm behind HiFi reads.

Iso-SeqStructural variantsConsensus
Oct 2015–Jul 2018Remote

Staff Engineer, Bioinformatics

Pacific Biosciences

Individual contributor developing statistical algorithms in C++14: polyploid consensus polishing, minor-variant reconstruction for personalized medicine, hi-plex demultiplexing with QC, and the pbmm2 alignment frontend for minimap2.

AlgorithmsC++pbmm2
Sep 2014–Sep 2015Remote

Senior Engineer, Bioinformatics

Pacific Biosciences

Hardware-near C++11 on x86_64 and MIC architectures for the Sequel instrument. Improved base-call accuracy and used hotspot parallelization and vectorization to enable real-time base calling. The starting point.

VectorizationSequelSIMD
2011–2014Basel, CH

Dr. sc., Computational Biology

ETH Zürich

Doctoral research on viral quasispecies reconstruction: statistical modeling of population structure from NGS and long-read data. Authored HaploClique and QuasiRecomb for maximal-clique quasispecies assembly and recombinant reconstruction.

Statistical modelingQuasispeciesGenomics
2009–2011Bielefeld, DE

M.Sc., Bioinformatics & Genome Research

University of Bielefeld

Thesis topic: Prediction of Group I Introns under structure variation.

BioinformaticsRNA structureGenome research
2006–2009Bielefeld, DE

B.Sc., Bioinformatics & Genome Research

University of Bielefeld

Thesis topic: Ideas for and implementation of an automated statistical data analysis.

BioinformaticsStatisticsGenome research

Skills · −890 m

What I bring to the table.

Leadership

Building and scaling teams that ship.

Org buildingCross-team executionRoadmap & OKRsExecutive communicationFully-remote teams

Domain & Algorithms

From raw signal to validated variants.

Long-read sequencingReal-time data pathsSignal processingConsensus callingVariant callingStatistical modelingV&V

Engineering & ML

High-performance, reproducible, GPU-accelerated.

C++23CUDASIMDCI/CDReproducible buildsONNXDL productizationMLOps

Selected research · −1050 m

Peer-reviewed, cited at scale.

2022

DeepConsensus improves the accuracy of sequences with a gap-aware sequence transformer

Nature Biotechnology
2019

Highly-accurate long-read sequencing improves variant detection and assembly of a human genome

Nature Biotechnology
2014

Full-length haplotype reconstruction to infer the structure of heterogeneous virus populations

Nucleic Acids Research

Open source · −1150 m

Shipped, and still in the field.

Contact · −1200 m

Let's talk.

Open to select opportunities in engineering leadership, genomics, and high-performance computing. The fastest way to reach me is email.

armin.toepfer@gmail.com