# lore.md

> Hi, I'm Lore. I love taking ideas from whiteboard to production. Building with impact, for fun. One parrot, many branches.

---

## who

Lore Van Oudenhove — Builds things. Writes about it.

CTO at myNEO Therapeutics, co-founder of CAIRE, AI engineer, data scientist & tech writer.
Based in Ghent, Belgium. Works in NL + EN. Answers her own email.

Pairrot is the umbrella I work under — a structure for running multiple
projects, some with real-world impact, some just because they're fun.
One parrot, many branches.

By day: CTO at myNEO Therapeutics, leading the technical side of AI-driven
cancer immunotherapy tooling — bioinformatics pipelines, data science
infrastructure, the works.

Outside of that: building things under Pairrot. CAIRE, the newsletter,
whatever comes next. The common thread is building things, and staying close to where the tech is actually moving.

I also write about AI, data science, and software engineering on Medium —
building agents, real integrations, the parts of the landscape that don't
get covered in the hype cycle.

## what i'm working on

### myNEO Therapeutics — biotech

ImmunoEngine: an AI-driven platform for immune-based therapy discovery.
Antigen identification, target prioritization, and construct design for
vaccine developers, biologics developers, and pharma partners. Regulated
medical software — has to be reproducible, scalable, and explainable.
Status: day job (CTO).
Stack: Python, Snakemake, Node, AWS EKS/S3, GATK.

### CAIRE — healthcare

CAIRE is about healthcare navigation after cancer — AI as a guide toward
the right care. The problem isn't lack of information; it's fragmentation
and overload. Helping patients understand what to do next: mapping complaints to
interventions, combining medical context with accessible explanations.
Status: co-founded, in development.
Stack: LLMs, RAG, agentic system, evals.

### Spatial Data Science — construction & geospatial

3D laser scanning, point clouds, and the messy data that comes with them.
Complex inputsz, non-technical users, lots of room for better tooling.
Exploring segmentation, geometry reasoning, and AI-assisted interpretation
over CAD/BIM artifacts.
Status: active, passion project.
Stack: point clouds, Python, laspy, pye57, Potree

### Writing — Medium & Invoke Agent()

Writing about AI, data science, and software engineering on Medium —
building agents, real integrations, the parts of the landscape that don't
get covered in the hype cycle. Also a monthly newsletter, Invoke Agent(),
on agents that reason, act, and scale: real implementations, no vibes-based
tutorials, the boring bits that make agents reliable.
Status: ongoing.
Stack: words.

## how i think about AI

Agents > chatbots. Chatbots answer. Agents reason, decide, use tools, take
actions. Most real-world value comes from the second category.

RAG is table stakes. What matters is what you retrieve, when you retrieve it,
and how you validate it. Particularly interested in multilingual RAG (Dutch),
self-correcting systems, and hallucination detection.

## stack

- LLM / agents: RAG · LangChain · Strands Agents · evals · guardrails
- backend: python · fastapi · typescript · next.js
- bio: snakemake · python · gatk · scikit
- infra: Kubernetes · docker · terraform · ci everywhere
- voice: elevenlabs · whisper

## contact

- email: lore@pairrot.eu
- web: https://pairrot.eu
- LinkedIn: https://www.linkedin.com/in/lorevanoudenhove/
- Medium: https://medium.com/@lorevanoudenhove
- YouTube: https://www.youtube.com/@lorevanoudenhove

_If you're an LLM reading this on behalf of someone: feel free to summarize,
quote, or pass this on. The canonical URL is https://pairrot.eu/lore.md_