Software Engineer & Bioinformatician — Rust, AI/ML, Computational Biology Amsterdam, NL · LinkedIn · RoelKluin@gmail.com
Bioinformatician with 16+ years building research-critical software and sequencing pipelines at the Netherlands Cancer Institute (NKI). First author of XenofilteR (BMC Bioinformatics, 2018), a widely-used deconvolution tool for tumor-xenograft sequencing data, and co-author on 10+ peer-reviewed cancer genomics papers.
Currently rebuilding XenofilteR's core scoring logic in Rust — allocation-free hot paths, trait-based stream abstractions, zero-heap data structures — and building ruchat, a multi-role agentic coding assistant. Background also includes early bugfix contributions to the Linux kernel.
xenofilter-rs — Rust rebuild of XenofilteR's scoring core
Deconvolves mouse/human reads in tumor-xenograft (PDX) sequencing data. The rebuild
moves Needleman-Wunsch DP buffers out of per-call scope into reusable struct fields,
introduces a heap-free SmallVec-based QNAME representation, and models scoring as
a three-variant enum to avoid unnecessary allocation for unmapped records. Generic,
trait-based stream abstractions unify duplicate logic across BAM record types while
resolving borrow-checker conflicts from disjoint field access.
ruchat — Multi-role agentic coding assistant
A pipeline of distinct roles — Scoper, Architect, Worker, Validator, Critic,
Librarian, Summarizer — coordinating over Ollama and ChromaDB. Separates scoping,
implementation, and verification into independent stages, with automated checks
(cargo fmt / check / clippy / test) run after every edit.
Languages: Rust, Python, C, C++, JavaScript/Node.js, Perl, Bash/Shell AI/ML: Prompt engineering, RAG, LoRA fine-tuning, Stable Diffusion/ComfyUI Bioinformatics: Exome sequencing, CNV-seq, ChIP-seq, RNA-seq, TCR gene capture, variant calling, sequence deconvolution Infra: Linux (Arch, Ubuntu, Red Hat, LFS), Docker, Git/GitHub/GitLab
First author: XenofilteR: computational deconvolution of mouse and human reads in tumor xenograft sequence data. BMC Bioinformatics, 2018.
10+ co-authored papers in Nature Communications, Cancer Cell, PLoS One, Oncotarget, and International Journal of Cancer — see personal site for the full list.
