2021 - Present
An open-source, self-hosted personal productivity platform built on a
microservices architecture. Loryth integrates task management, a knowledge
base, grocery planning, journaling, and AI assistance into a unified
system — designed for users who want full ownership of their personal data
without relying on fragmented corporate services.
Mar 2023 - Aug 2023 (5 months)
An open-source ride-sharing platform for Greece, named after the Straw Hat
Pirates' first ship in One Piece. Inspired by the 2023 Tempi rail disaster,
which highlighted the lack of reliable alternative transportation options in
Greece, Going Merry was built to connect drivers and passengers for
long-distance rides with a focus on privacy and trust.
Feb 2024 (weekend project)
A PyCharm plugin that fixes type resolution for SQLAlchemy 2.0's Mapped[T]
annotations. PyCharm incorrectly flags valid SQLAlchemy code as type errors
— this plugin hooks into the IDE's type provider system to correctly resolve
mapped attributes on both instances and classes, supporting features like
@hybrid_property and @declared_attr. A weekend hack that picked up 20 stars
on GitHub.
An alternative executor for TensorFlow that replaces the default breadth-first
(eager) graph execution with a demand-driven, depth-first approach. By propagating
demand backward from output nodes, intermediate results are consumed sooner and
freed earlier — dramatically reducing peak memory usage during operations like
backpropagation, where standard execution forces early activations to persist
until the very end of the backward pass. Developed as an undergraduate research
project at the University of Athens, with a working prototype showing promising
memory reduction results.
Represented the University of Athens in the ACM SIGMOD 2020 Programming Contest,
an entity resolution challenge requiring participants to match duplicate product
specifications across 24 e-commerce websites. The system needed to identify which
of ~30,000 camera product listings referred to the same real-world product,
evaluated by F-measure on a secret held-out dataset.