Amplifier AI is building a next-generation surgical planning platform - powered by our own imaging, segmentation, and 3D measurement engine.
Weβre a small, no-bullshit team of data scientists, radiologists, developers, and serial entrepreneurs with successful exits - from Ukraine and across Europe. We care about precision, reproducibility, and building tools that actually help doctors plan safer surgeries.
If you love clean Python, 3D geometry, and the smell of medical DICOMs in the morning - read on.
Why Amplifier
Real impact: Your work will directly improve surgical planning and patient outcomes.
Mentorship: A rare chance to dive deep into medical imaging while apprenticing with the CTO.
Growth: Competitive salary and opportunity to get stock options.
Team: A thoughtful, senior, multinational crew of engineers, data scientists, and clinicians - all united by curiosity and precision.
How We Work
Remote: Yes
Time: Flexible, with EU overlap
Communication: Ukrainian, English
Mindset: Precision, curiosity, no corporate BS
And letβs be honest:
- Thereβs micromanagement - not out of control, but you will get code reviews that feel like philosophy exams.
- Deadlines are brutal. We ship fast because hospitals and surgeries donβt wait.
- Priorities may change overnight. Clinical partners and real-world data tend to break plans.
But every time you push a fix that makes a surgery safer, you remember why itβs worth it.
Weβre not building βAI for fun.β Weβre helping doctors save lives.
Welcome to the hellishly good team.
What Youβll Work On
Youβll join the team behind our internal Python library that powers all 3D reconstruction and measurement logic for CT-based clinical workflows.
Your playground includes:
- SimpleITK & DICOM: Reading DICOM series, validating LPS origins/directions, resampling safely, and maintaining voxel β physical transforms.
- VTK & Geometry: Generating surfaces from labels, computing distances, intersections, and clipping; exporting nearest-point markers and reliable STL files.
- Measurement & QA logic: Building robust procedural tools that make radiologistsβ lives easier (and their reports cleaner).
- Pipeline reliability: Tuning performance, adding structured logging, and ensuring everything survives real hospital data.
Our Stack
Core: Python 3.10+, SimpleITK, VTK, NumPy/SciPy, Pydantic
ML: PyTorch, nnU-Net (for segmentation & experiments)
Tooling: pytest, mypy (strict), ruff, pre-commit, Docker, GitHub Actions, Git LFS, Sentry
We write clean, typed code. Reproducibility matters. And we prefer simple, readable solutions over βcleverβ ones.
What Weβre Looking For
Weβre not hiring for a specific seniority level.
Weβre hiring thinkers - people who love solving problems and arenβt afraid to go deep into geometry, image transforms, and pixel-level precision.
Youβll thrive if you:
- Write clean, modular Python code.
- Understand how DICOMs work in 3D - origin, spacing, direction, resampling.
- Know the difference between voxel and physical space, and how to transform between them.
- Enjoy debugging weird edge cases and care about correctness.
- Can explain things clearly in English (we love docstrings and discussions).
If youβve worked with SimpleITK, VTK, or PyTorch, thatβs a strong plus.
If not - weβll teach you. Weβve built internal training pipelines and mentoring for people who think fast and learn faster.
Nice to Have
- Experience with VTK surface/volume rendering or pipeline optimization.
- Background in ML segmentation (Dice, Hausdorff, nnU-Net).
- Curiosity about GPU acceleration and large-volume performance.
Example Challenges
- Build a robust distance-measurement tool between anatomical segmentations, with exportable nearest-point STL markers.
- Extract centerlines and landmarks from noisy CT scans.
- Enforce consistent spacing/origin/direction when combining data from different scanners and protocols.
If these sound fun rather than scary, youβll fit right in.
How to Apply
Send us:
- Your CV or LinkedIn
- Your GitHub or portfolio (we care more about code than titles)
- A few lines about what youβve built with Python, and - if youβve touched SimpleITK or VTK - tell us briefly how.
A Note from Us
Weβre Ukrainian engineers building something that blends science and purpose.
Our tools help radiologists and surgeons plan procedures that save lives.
We donβt do hype. We build things that work - and weβll help you grow into an expert if you bring curiosity and grit.