Hi, my name is Piotr Pasza Storożenko located in Warsaw, Poland. I’m a machine learning engineer, physicist, by heart with love for programming. Wannabe applied mathematician. A huge julia enthusiast, but I code mostly in python and R. Some small chunks of code and simulations in C++17 and newer are also fine ^^. On this blog I share things I find interesting.

Experience

  • Fall 2023 - Now – Innovation Lead@Appsilon

    Who’s the Innovation Lead? An experienced engineer outside of regular projects who helps to make company future-ready. Glue work is my work. I’m the go-to person that connects the tech team, delivery, marketing, sales, supports the board when it comes to new technologies, architectures, and methodologies, or innovation to put it simply. In the meantime, I try to code, below are some of the projects and initiatives I am or was involved in:

    • ML Systems Development
      • Architected and deployed image classification ML system using BentoML (FastAPI + Docker) on GCP Cloud Run, implementing cost-efficient serverless infrastructure
      • Developed high-performance data visualization dashboard for ML research, handling a lot of data points and contributing to a Nature publication
      • Evaluated and implemented Nextflow for ML workflow management, resulting in successful client adoption and multiple project extensions
    • Technical Leadership
      • Built LLM-powered application Flow, this is like customizable Claude Artifact, but for RShiny apps, Quarto reports and other applications.
      • Created an open-source Tapyr package for Shiny for Python application bootstrapping
      • Created full-stack architecture connecting ML models with modern frontend frameworks (React, Next.js, Observable Framework)
      • The main code reviewer for python code in the company.
      • I brought the torch of knowledge to our python devs by introducing them to uv (earlier to poetry), ruff, pyright (earlier mypy), playwright, pydantic-settings, hydra, and other modern tools.
    • Strategic Innovation
      • Led technology evaluation initiatives identifying alternatives to core R Shiny business, resulting in expanded service offerings
      • Established structured innovation process balancing exploratory research with client-driven development
      • Bridged gap between technical capabilities and market needs, translating POCs into commercial engagements
  • 2021 - Fall 2023 – ML Engineer@Appsilon

    • Research in 3D animation modeling using unusual GANs - GenAI before it was cool.
      • Tech lead behind porting student-grade pytorch code to production-grade pytorch lightning with tests, typing, CI/CD, model serialization, deployment, and more.
      • Designed and implemented architecture for robust data validation
    • Research in artistic sketch to photorealistic 3D model area. Working with SoTA GANs models.
      • Explorative and innovative research that can be summarized as Take a few (over 20) different GANs, classifiers, algorithms, and mix them into pipeline that will generate riggable head 3D models from sketches.
      • Apart from ML engineering parts of the pipeline, I was the tech lead behind orchestrating all parts of the pipeline.
      • Everything in Pytorch, FastAPI, GCP stack.
    • Counting Nests of Shags in Antarctic with IBB PAS
      • YOLOv6-based model for counting nests of shags in Antarctica.
      • Co-invented and implemented a novel algorithm for counting objects from image patches. See Inference Method in paper for more details.
    • Collaboration with professor Frédéric Maps from Université Laval on detecting oil sacs in copepods - real-world image segmentation problem.
    • Improvements in Mbaza project - production classification models for camera traps, deployed in Gabon. 💙💙💙
  • 2020 - 2021 – Data Scientist@ING Bank Śląski

    • Part of ING Poland Center of Expertise - Artificial Intelligence. Data Science consulting for various internal teams. Both problem understanding as well as delivering quick PoCs.
    • Creating Polish language NLP model for sentence multi-label classification, Corporate Clients Department. Mixing various techniques of transformers with classic ML.
    • Creating and teaching at internal AI Academy - explaining basics of data processing and ML in python for over 100 people in a few 8-week courses.
    • Working on the whole company standards for machine learning.
  • Summer 2019 – Research internship SURF@IFISC, UIB, Palma de Mallorca

    • Agent based model of Twitter hashtags dynamics in numpy, cython and eventually C++17.
    • Analyzing 273 GB of tweets in python and R.
    • Internship expanded into Master thesis later on.
  • Summer 2017 – Internship in Advanced NLP Group@Samsung R&D

    • Working on the next word prediction engine for Samsung Galaxy’s keyboard, mostly C++.
  • 2016 - 2020 – Science Teacher@Inceptio

    • Around ~10 6-weeks courses for high-school students on C++, java, python.
    • Caretaker and teacher at WUT Summer Camps 2017-2020. Courses on Data Science, Physics, Math, Programming and more for high school students.

Other

Education

  • 2018 - 2020 Master in Data Science, MiNI, Warsaw University of Technology, thesis title Competition for attention in online social networks under dr inż. Grzegorz Siudem from WUT as well as dr José Javier Ramasco and dr Sandro Meloni from IFISC, summa cum laude.
    • For the first 2 semesters, I studied Computer Science which gave me a solid foundation in algorithms, data structures, and programming, and then I switched to the dedicated Data Science program.
    • Winter 2019 Erasmus exchange, Faculty of Engineering Science, KU Leuven
    • I was honored to give a graduation speech.
  • 2014 - 2018 Bachelor of Eng in Applied Physics, Faculty of Physics, Warsaw University of Technology, thesis title Synthesis and cristalization of glass and nanomaterials LiFe0.85V0.1PO4 under dr inż. Tomasz Pietrzak, summa cum laude