david

Hi, I’m David

I like getting deep into a subject and building something with what I learn.

I’m a software & ML engineer at Apple in Munich. Before that, I researched brain surgery imaging at Harvard Medical School and led TUM.ai.

My interests tend to wander into biology, physics, and how models learn.

Outside of all that: a lot of pickleball.
(Padel, wait for me 👀)

Say helloGitHubLinkedInCV
David taking a mirror selfie
David Podolskyi
Posing beside a Fallout Vault Boy statue
also me, apparently

A few things I’ve worked on

All projects
Rare Candy Lab 🍬

Teaching a model to play Pokémon

A game-playing agent built from scratch: learn from games, play against itself, then search ahead. Top 5% on Kaggle.

Two CPUs at match time. No GPU or internet.

Harvard Medical School / Bachelor’s thesis

Neural rendering for brain surgery

Matching a surgeon’s camera view to brain scans using neural radiance fields. My Bachelor’s thesis at Harvard Medical School.

Research in image-guided neurosurgery.

Apple / Software & machine learning

From a software issue to a code fix

At Apple, I built a system that takes internal software issues through to code changes ready for review, and train the models that do the work.

Built and shipped the production system.

EDTH Paris Hackathon / December 2024 / 2nd place

Magnetic navigation for submarine drones

I led implementation of an iPhone-based magnetic and inertial navigation prototype, from sensor capture to a working indoor demonstration. Our team won 2nd place at EDTH Paris.

2nd place. Demonstrated indoors at Station F.

The TUM.ai team together in matching hoodies
Some of the people who made TUM.ai what it is.

A lot happened at TUM.ai.

I joined early and later spent two years as president. We grew to 300+ members, ran summits and hackathons, and worked with students on research and startup ideas.

I’m proud of what we built together. There are a lot of people behind that sentence.

More about that part of my life

Where I’ve spent my time

Download CV

Work

  1. Apple

    Aug 2026 – present

    Software & Machine Learning Engineer

  2. Apple

    Dec 2025 – Aug 2026

    Software & ML Engineer Intern

  3. UnternehmerTUM, Venture Labs Software/AI

    Nov 2024 – Jun 2025

    Project Manager

  4. Harvard Medical School

    Apr – Sep 2024

    Machine Learning Researcher, Brain Surgery Imaging

  5. AISBACH

    Feb – Jun 2024

    ML & Software Engineering Consultant

  6. Infineon Technologies

    Sep 2022 – Jan 2024

    AI Solutions Engineer

  7. Expleo Group

    Mar 2021 – Sep 2022

    Data Engineer for IoT Systems

  8. TURTLE GmbH

    Mar – Jun 2022

    Data Engineer & Science Consultant

  9. AI Driller

    Aug 2020 – Jan 2021

    Junior Data Scientist & ML Engineer

Volunteering & community

  1. TUM.ai

    Sep 2022 – Sep 2024

    President

  2. TUM.ai

    Nov 2021 – Sep 2022

    Industry Department, Team Lead & Mentor

  3. UnternehmerTUM & AppliedAI

    Mar – Apr 2022

    Help hotline for Ukrainian refugees

  4. TUM.ai

    Jun 2021 – Nov 2022

    Venture Department, Team Lead

  5. TUM.ai

    Nov 2020 – Jun 2021

    Events Organiser & ML Consultant

On my mind lately

What actually happens inside a model as it learns? Why can training it to do one thing make it worse at something else? I want to work on explainable AI, especially mechanistic interpretability, to understand those changes.

AI alignment is another question I want to work on: how do we build models whose behaviour we can trust, even in situations they weren’t trained for?

I’m also curious about where this meets biology, and what we could learn by looking inside models trained on biological data.

And I want to keep finding ways to use what I know to help Ukraine.

Play with the code.