PyCon Lithuania: 15 Editions, 2009–2026

2009 — The Beginning

April 26 | KTU Gymnasium, Studentų g. 65, Kaunas | ~30 attendees Organisers: Jurgis Pralgauskis, TBD. Marius Gedminas built the minimalist conference website.

14 talks, one day, a handful of people who cared enough to show up and share what they knew.

The programme: web2py and Plone got full talks, Django wasn't even mentioned. WSGI needed explaining. zc.buildout was the serious deployment tool. Mercurial got a talk. Python on a mobile phone was a frontier topic. NLP was squeezed into the last slot of the day.

Jurgis gave five talks himself. Albertas Agejevas three. Someone from the audience jumped in to co-present. No sponsors. Just community.

Sponsors: — Further: pykonf.albinas.lt/2009 (videos)

In hindsight: A time capsule of Python at a crossroads. The web framework wars were still undecided. pip and virtualenv hadn't become the default yet. Git hadn't won. And NLP — almost an afterthought in the last slot — is sixteen years later the reason half the world is talking about Python.

2010–2011 — The Gap

No conference. Three years of silence between the first and second editions.

2012 — Regrouping in Vilnius

April 28 | VIKO College, J. Jasinskio g. 15, Vilnius | ~40 attendees Organisers: Mantas Zimnickas, TBD

The community regrouped and moved to the capital. Seven talks, all recorded and still on YouTube.

The topics — Fabric, Python packaging, Android, Google App Engine, HPC, teaching Python, Kivy — reflect practitioners solving real problems. There were mobile and GUI talks (Android, Kivy), and an HPC talk from Vilnius University.

Sponsors: — Further: YouTube playlist (videos) | web.archive.org (archive)

In hindsight: Not a single mention of machine learning or data science. This was two years before Python's ascent to AI dominance. Nobody saw it coming. The mobile optimism didn't materialise — Python never conquered mobile. But the HPC talk quietly pointed toward the scientific Python wave.

2013 — Finding Its Own Space

April 27 | Hub Vilnius, Šermukšnių 6A, Vilnius | ~50 attendees Organisers: Mantas Zimnickas. Domantas Jackūnas helped organise and moderated.

From a university classroom to a coworking hub. The community was small but finding its own space — literally.

Domantas Jackūnas gave "Nuo Django prie Flask." SaltStack got a full session. Mantas Zimnickas talked about pandas at version 0.10. Albertas Gimbutas gave two back-to-back talks — OpenERP and OpenMPI. Enterprise resource planning and parallel computing from the same person in one afternoon. That's the energy of a small conference.

Jurgis Pralgauskis, who started it all in 2009, was still presenting.

Sponsors: — Further: GitHub (source)

In hindsight: The Django-to-Flask migration story thousands would live through as microframeworks gained ground. Infrastructure-as-code was becoming non-optional. And pandas — still finding its audience — would become synonymous with Python data work within a year.

2014 — The Hiring Signal

May 10 | VU MIF, Naugarduko g. 24, Vilnius | ~50 attendees Organisers: Mantas Zimnickas, Albertas Gimbutas, Laurynas Speičys. Albertas moderated. Laurynas handled finances.

The conference moved to Vilnius University's Faculty of Mathematics and Informatics — and stayed for three years. It opened with something new: a hiring session. The Python job market in Lithuania was real enough for recruiters to show up before the first talk.

Marius Gedminas kicked off with "Plepalai apie smauglį" (Chats about the python). Mantas Zimnickas talked "Apie sūrį ir baterijas" (cheese and batteries) — packaging (CheeseShop, the old PyPI name) and the standard library. pip had just become the recommended installer. The wheel format was brand new.

The lightning talks had surprises: Mantas on Python for Lithuanian language analysis (NLP for a language of 3 million speakers), Jurgis on a Python-to-other-languages transpiler, and Remigijus on logging and decorators back-to-back.

Sponsors: — Further: YouTube playlist (videos) | GitHub (source)

In hindsight: NLP for Lithuanian — years before multilingual models made small-language processing easier. And pip install still didn't just work reliably on all platforms. It would take years.

2015 — Typing Before It Was Cool

April 25 | VU MIF, Naugarduko g. 24, Room 103, Vilnius | ~60 attendees | Free admission Organisers: Mantas Zimnickas, Laurynas Speičys, Justas Trimailovas, Albertas Gimbutas. Justas made t-shirts and attendee badges. Mantas built the website with a new design. Laurynas moderated and helped with finances.

Ignas Vyšniauskas gave a talk on type hints and mypy — months before PEP 484 was accepted. The idea that Python could have optional static types was still controversial.

Justas Trimailovas talked Python on microcontrollers. MicroPython had been released just a year earlier. Rokas Aleksiūnas presented on Apache Mesos — serious backing from Twitter and Airbnb.

And in the lightning talks, a name appeared for the first time: Aidis Stukas, asking "Kada mokytis Python jau per vėlu?" (When is it too late to learn Python?)

Sponsors: Programmers of Vilnius, Uber, hbee Further: —

In hindsight: A decade later, type hints are everywhere and mypy is an industry standard. Ignas was early. Kubernetes would win decisively within two years and Mesos would fade — conference talks are sometimes time capsules of paths not taken. And Aidis Stukas would go on to become one of the conference organisers.

2016 — Async Awakening

May 7 | VU MIF, Naugarduko g. 24, Room 103, Vilnius | ~80 attendees | Free admission Organisers: Mantas Zimnickas, Laurynas Speičys, Justas Trimailovas

Two async/real-time talks back to back: Django Channels (Justas Trimailovas) and crossbar.io (Justas Sadževičius). The community was hitting the limits of request-response.

Rokas Aleksiūnas, who talked Mesos the year before, now talked microservices. Petras Zdanavičius built a LISP interpreter in Python PLY — the kind of talk that reminds you conferences aren't just about production tools. Laurynas Speičys presented data on Python jobs in Lithuania. The market was growing and the community was tracking it.

Sponsors: Revel Systems (Gold, 1024 EUR), Strategic Staffing Solutions & Programmers of Vilnius (Silver), hbee & Asteris (Bronze) Further: —

In hindsight: Django Channels was the first serious attempt at async Django — native support wouldn't land until Django 3.0 in 2019. Microservices were peaking on the hype cycle. Everyone was decomposing monoliths. The industry would spend the next five years learning which ones should have stayed whole.

2017 — The Leap

May 13 | KTU Santaka Valley, K. Baršausko g. 59, Kaunas | ~120 attendees Organisers: Aidis Stukas, Jurgis Pralgauskis, Mantas Zimnickas, Laurynas Speičys, Andrejus TBD

Everything changed. For the first time: two parallel tracks. Hands-on workshops (Python basics, Django, pandas, scikit-learn). And the first international speaker — Andrew Svetlov, a CPython core developer and aiohttp maintainer from Ukraine, who gave two talks: building open source libraries and optimising Python with Cython.

The conference switched to English. Eight years of Lithuanian-language talks — "Plepalai apie smauglį," "Kada mokytis Python jau per vėlu?" — gave way to an international format. The community was no longer just talking to itself.

Rivo Laks came from Estonia to talk Django + React. Simona Bekėraitė talked Python in astrophysics. Mantas Zimnickas gave "import antigravity" — a fitting title for the year PyCon LT started to fly.

Sponsors: Revel Systems, Versada, Programmers of Vilnius, KTU Startup Space, Strategic Staffing Solutions | Media: JetBrains, O'Reilly, PythonAnywhere Further: —

In hindsight: Andrew Svetlov came from Ukraine five years before Russia's full-scale invasion made that origin carry a different weight. The Django + React split Rivo Laks described would define web development for the next decade. And the scikit-learn workshop introduced a tool that was about to become the entry point for an entire generation of ML practitioners.

2018 — Kaunas Again

May 5 | KTU, K. Baršausko g. 59, Kaunas | ~150 attendees Organisers: Aidis Stukas, Agnija Ravdo, Karina TBD, Laurynas Speičys, Jurgis Pralgauskis, Džiugas Tornau, Dovilė Urbanavičiūtė

The second year in Kaunas. Two parallel tracks — one PyData, one general — and the first dedicated keynote: Mike Müller, long-time Python trainer and PSF Fellow.

The PyData track was growing teeth: high-performance pandas in economics (Pietro Battiston), text analysis (Bhargav), deploying ML models (Sebastian Neubauer), Jupyter as a web application platform (Nikolay Karelin). A DataDog speaker and an AdForm speaker gave company-backed talks — industry was investing in the conference.

The general track had Antonio Cuni on PyPy, Dmitry Nazarov on ASGI and Daphne (the async web future, again), Angela Branaes on building a startup with Python and React, and Šarūnas Navickas on monitoring. Petras Zdanavičius was still presenting.

Sponsors: TBD Further: Hall 1 stream | Hall 2 stream (videos)

In hindsight: ASGI and Daphne — the async web story continued from Django Channels in 2016. It would take until Django 3.0 and then Starlette and FastAPI to deliver on what these talks kept promising. The PyData track's arrival as a first-class citizen foreshadowed the themed days that would come in 2024.

2019 — The Inflection Point

May 25–26 | SMK (University of Social Sciences), Vilnius | ~350 attendees Organisers: Aidis Stukas, Inga Pliavgo, Dovilė Urbanavičiūtė

The first two-day PyCon Lithuania. Back in Vilnius, and the attendance more than doubled — roughly 350 people, up from ~150 the year before. This was the growth inflection point.

The speaker list told you this wasn't a local meetup anymore. Three keynotes: Ian Ozsvald (author of "High Performance Python") on citizen science, John Bywater on events as the modern object orientation, and Hynek Schlawack — maintainer of attrs, structlog, and a fixture of the European Python circuit — on maintaining a Python project when it's not your job. Radovan Kavický surveyed data science past, present and future.

The data science track had arrived in force: production Spark workloads (Chetan Khatri), ML from notebook to production (Matas Šeimys), data pipelines (Oleg Shydlouski), geodata with JupyterHub (Martin Christen), emotion detection from video (Justin Shenk). Inga Popovaitė used Python to study group behaviour in space analog facilities.

Sponsors: TBD Further: YouTube playlist (videos)

In hindsight: They had no idea what was coming. The momentum they'd built — 350 people, two days, international speakers — would be frozen for almost two years. The 2020 edition was already being planned.

2020 — COVID Cancellation

No conference. The pandemic hit weeks before the planned event. Tickets had already been on sale.

In hindsight: The gap lasted two editions' worth of momentum. When the conference returned in 2021, attendance had halved. It took until 2022 to get back to 350. But the forced pause may have helped — the organisers came back with bigger ambitions, not smaller ones.

2021 — The Return

September 3 | Best Western, Konstitucijos pr. 14, Vilnius | ~150 attendees Organisers: Aidis Stukas, Inga Pliavgo, Dainora Bučytė

The post-COVID return — in September, not the usual spring slot. Galimybių Pasas (Lithuania's National Certificate) required for entry. Masks mandatory. Deliberately scaled down to reduce risk while COVID was still unfolding. The community showed up anyway.

Sponsors: TBD Further: —

In hindsight: The deliberate scale-down worked — a small, safe return that proved the community was still there. Within a year they'd be back at 350. Within three, at 600+.

2022 — The International Roster

May 26–27 | SMK, Vilnius | ~350 attendees Organisers: Aidis Stukas, Inga Pliavgo, ...

Two days, 30 talks across four tracks — Python Room, PyData Room, Education Summit, and keynotes in the Main Room — and a speaker list that would have been unthinkable five years earlier.

Sebastián Ramírez (tiangolo), creator of FastAPI, talked about mixing async and blocking code. Łukasz Langa, CPython Developer in Residence and creator of Black, showed how typing speeds up apps with mypyc. James Powell delivered his characteristically deep "int is to list as float is to tuple."

The ML/ops track was growing fast: MLOps with Kubernetes, deployment challenges, user-centric machine learning. Cheuk Ting Ho warned about Trojan Source Code. Juan Luis Cano Rodríguez staged "The great Python dataframe showdown."

And the local thread kept running. Albertas Gimbutas, who gave two talks on OpenERP and OpenMPI back in 2013, was now scaling WebSockets. Petras Zdanavičius, who'd been recording talks since 2012, was implementing ray tracing in Python. Aidis Stukas — who once asked "Kada mokytis Python jau per vėlu?" — was now giving the opening introduction as organiser. Rokas Cvirka showed up with three sessions. The spirit of Jurgis giving five talks in 2009 was alive.

Sponsors: TBD Further: Pretalx

In hindsight: The dataframe showdown signalled what was coming — Polars, DuckDB, Arrow-backed everything. The post-pandas world was taking shape, though most teams hadn't noticed yet.

2023 — The Creators Show Up

May 17–20 | Crowne Plaza Hotel, Vilnius | ~450 attendees Organisers: Aidis Stukas, Inga Pliavgo, ...

The first four-day edition: tutorials, two conference days, sprints. 42 talks. The move to the Crowne Plaza wasn't planned ambition — SMK, the venue for 2019 and 2022, didn't honour a verbal agreement, and the organisers had to scramble for a new space. They landed in a hotel and made it work. PyCon Lithuania was now billing itself as the biggest Python and PyData event in the Baltics and Nordics.

The headliner list read like a who's-who of the Python data stack. Travis Oliphant, creator of NumPy and SciPy, gave the keynote and a talk on building sustainable AI/ML software. Samuel Colvin, creator of Pydantic, delivered "Garbage in → Pydantic → you're golden!" Ritchie Vink, creator of Polars, presented "done the fast, now the scale." Marc Garcia (pandas core) gave three sessions — pandas 2.0 and the Arrow revolution, first open source contribution, and an Uncle Data panel with Vink.

Marlene Mhangami (former PSF Director) spoke on Python and Creativity. Dr. Thomas Wiecki (PyMC Labs) brought Bayesian methods to business. Sanskar Jethi introduced Robyn — a Python web framework with a Rust runtime. The data/ML track was now dominant: Streamlit + WebAssembly, feature engineering with Hamilton, billion-scale vector search, recommender systems, purchase prediction in mobile games.

The pattern was clear: the people building the tools Python developers use every day were now coming to Vilnius to talk about them.

Sponsors: TBD Further: Pretalx

In hindsight: "I talk to ChatGPT about things" was a conference talk — ChatGPT had been public for six months. Within a year it would reshape how every developer in the room worked. Pydantic v2 (rewritten in Rust) shipped that summer. The Arrow revolution Marc Garcia described did happen — and Polars went from "interesting alternative" to essential tooling. Robyn's Rust runtime was an early signal of the Rust-in-Python pattern that would accelerate through PyO3, Pydantic, Ruff, and uv.

2024 — Themed Days and 600+

April 2–6 | SMK, Kalvarijų g. 137E, Vilnius | 600+ attendees Organisers: Aidis Stukas, Inga Pliavgo, ...

The conference hit 600 attendees and introduced themed days: Web Day, Python Day, Data Day, plus tutorials and sprints. Three parallel tracks. 65 talks.

Six keynotes from people who shaped the ecosystem: Tom Christie (Django REST Framework, Starlette, httpx), Daniel Roy Greenfeld ("Two Scoops of Django"), Arjan Egges (ArjanCodes), Robert Smallshire (Python Apprentice/Journeyman/Master), Ines Montani (spaCy, Explosion), Ritchie Vink (Polars, returning for a second year).

The depth was new. Marco Gorelli (Polars core) gave both a tutorial and a talk on DataFrame interoperability. Rodrigo Girão Serrão shared lessons from 503 days of full-time FOSS work. Tom Christie talked about funding open source development — a topic the community rarely confronts honestly. Matthew Honnibal (spaCy co-creator) looked at tomorrow's programming workflows. The Rust-Python bridge got a dedicated session: David Hewitt on PyO3 and how it made Pydantic v2 fast.

Three Maxim Danilov talks in one conference — µDjango 2.0, Django FTL, and distributing complexity. The multi-talk-per-speaker tradition from 2009 was still going.

Sponsors: Oxylabs (Diamond), Gurtam (Gold), Carvertical, CornerCase Technologies, Shift4 (Silver), Revel Systems, AppSignal (Bronze) Further: Pretalx | Tickets: Individual 69/99/129 EUR, Corporate 99/139/179 EUR (1/2/3 days)

In hindsight: Ines Montani's "The AI Revolution Will Not Be Monopolized" was prescient — open-source models would close the gap with proprietary ones faster than most predicted. "Is Mojo Just a Hype?" — the answer, so far, is yes. And Matthew Honnibal's talk on tomorrow's programming workflows landed months before AI coding assistants changed the daily experience of writing software.

2025 — AI Gets Its Own Day

April 23–25 + Sprints April 26 | SMK, Kalvarijų g. 137E, Vilnius | 600+ attendees Organisers: Aidis Stukas, Inga Pliavgo, ...

Three themed days: Python Day, Data Day, AI & ML Day. The dedicated AI day reflected the industry shift. 72 confirmed talks.

Six keynotes: Kushal Das (CPython core, Tor Project, PSF Director), Brandon Rhodes (legendary PyCon speaker), Stefanie Molin (Pandas author, Bloomberg), Gabor Szarnyas (DuckDB), Merve Noyan (Hugging Face), Aurimas Griciūnas (SwirlAI, AI infrastructure).

The Python Day had Rhodes on "Skip the Design Patterns: Architecting with Nouns and Verbs" and Das on ethics and privacy — the kind of pairing that only happens when a conference has range. Django was still going strong: async coroutines, pure REST APIs. The GIL's future got a full session. Virtual environments got an anatomy lesson.

The AI & ML Day was the new centre of gravity: Merve Noyan on open-source multimodal AI, agentic systems and workflows, knowledge bases for AI agents, RAG improvements, LangFlow, MCP with dynamic tool discovery, LLM fairness evaluation. The vocabulary had shifted completely from even two years earlier.

And a Lithuanian company outperformed Microsoft, Google, and OpenAI in speech-to-text. Alius Petraška told that story on Data Day.

Each conference day had a named sponsor: Oxylabs (Python), Carvertical (Data), Revolut (AI). PyLadies expanded with a dedicated lunch and a full workshop.

Sponsors: Oxylabs (Python Day), Carvertical (Data Day), Revolut (AI Day) Further: Pretalx

In hindsight: The vocabulary on AI Day — agentic systems, RAG, MCP, tool discovery, knowledge bases — was less than a year old. Half of these terms didn't exist at the 2023 edition. Whether they'll age like "type hints" (prescient) or "Apache Mesos" (a path not taken) is the question the 2025 programme leaves open.

2026 — 15th Edition

April 8–10 | Vilnius | Target: 600 attendees Organisers: Aidis Stukas, Inga Pliavgo, ...

From a handful of people in a Kaunas classroom to the largest Python event in the Baltics. From one person giving five talks to 72 sessions with international keynotes. From free admission in university rooms to a three-day professional conference.

The part that hasn't changed: a community figuring things out together.

Sponsors: TBD Further: pycon.lt/2026/tickets | Tickets: Individual 99/139/169 EUR, Corporate 139/169/249 EUR (1/2/3 days)

The Arc

Attendee numbers for 2009–2018 are estimates based on venue sizes and community accounts.

The Speakers

Creators, core developers, and authors who came to Vilnius (and Kaunas) over the years.

From a CPython core developer in a Kaunas lecture hall (2017) to the creators of FastAPI, Pydantic, Polars, NumPy, spaCy, and Django REST Framework on stage in Vilnius. The conference grew into its speaker list.