yiannisha Github
STATS3 FAILED STARTUPS0 SUCCESS STORIESCOUNTLESS MORE FAILURESCURRENT OBSESSIONSNNs, CUDA, RUST, LLMs, GPU Architecture, STATSTHINGS IM GOOD AT-THINGS I HAVE QUESTIONABLE EXPERIENCE INFULL-STACK
(REACT, NEXTJS, FLASK, DJANGO, NODEJS)CUML, SCIKIT, XGBOOST, PYTORCHRUSTSOLIDITY (NOT INTO CRYPTO ANYMORE)
STATS3 FAILED STARTUPS0 SUCCESS STORIESCOUNTLESS MORE FAILURESCURRENT OBSESSIONSNNs, CUDA, RUST, LLMs, GPU Architecture, STATSTHINGS IM GOOD AT-THINGS I HAVE QUESTIONABLE EXPERIENCE INFULL-STACK
(REACT, NEXTJS, FLASK, DJANGO, NODEJS)CUML, SCIKIT, XGBOOST, PYTORCHRUSTSOLIDITY (NOT INTO CRYPTO ANYMORE)
Mediocre generalist software engineer, with some experience in hardware stuff as well.
I like writing code and training shitty models in my free time.
I also do it as a job sometimes.
A bit of my story:
Started programming in high school, mostly gamedev stuff.
Did ~14 months of military service as an EOD, learned python and got my first freelance programming job in the meantime.
Managed to get into the ECE School of NTUA doing a MEng.
After 1.5 year of studying I dropped out to work full-time in different startups including my own.
In the span of 2 years I managed to work with 4-5 different startups, and co-founding 2 of them.
I worked in energy (pv), finance, blockchain, manufacturing, computer vision, and AI projects.
I was basically the tech lead in every single project I worked on.
At that time working ~12-14 hours a day was the norm. I even had a weekly 18-20 hour session on Fridays.
Eventually I realised that the reason I was doing all these stuff was more out of curiosity and the need to learn new things rather than the money/lifestyle.
I LOVE learning new things and solving problems.
I truly believe that this is the reason I am on this planet.
So, I eventually converged onto a single project, NAS-AI which is by far my favourite project I EVER worked on.
But I quickly ended up doing a lot non-technical stuff, which I loved but they were an obstacle to my learning.
At that time I realised that in order to do what I value most - learning and solving problems - at its highest level, I would need to get back to the basics and focus more on learning than on doing, at least for a while.
So, now I'm back in uni, finishing my MEng and working part-time in a startup that I love.
Everything that I learned during my time working actually helped me a lot in my studies, to the point that I'm gonna graduate around the time that I would have graduated if I didn't drop out.
I also consider doing a PhD once I graduate.
Currently working as a SWE at HeyFIRE
Also trying to graduate NTUA.
On my way to break into the AI research / MLOps field.
Get in touch:
me@yiannisha.dev
Some projects I've worked on:
Startups I cofounded:
Places I worked at:
I like writing code and training shitty models in my free time.
I also do it as a job sometimes.
A bit of my story:
Started programming in high school, mostly gamedev stuff.
Did ~14 months of military service as an EOD, learned python and got my first freelance programming job in the meantime.
Managed to get into the ECE School of NTUA doing a MEng.
After 1.5 year of studying I dropped out to work full-time in different startups including my own.
In the span of 2 years I managed to work with 4-5 different startups, and co-founding 2 of them.
I worked in energy (pv), finance, blockchain, manufacturing, computer vision, and AI projects.
I was basically the tech lead in every single project I worked on.
At that time working ~12-14 hours a day was the norm. I even had a weekly 18-20 hour session on Fridays.
Eventually I realised that the reason I was doing all these stuff was more out of curiosity and the need to learn new things rather than the money/lifestyle.
I LOVE learning new things and solving problems.
I truly believe that this is the reason I am on this planet.
So, I eventually converged onto a single project, NAS-AI which is by far my favourite project I EVER worked on.
But I quickly ended up doing a lot non-technical stuff, which I loved but they were an obstacle to my learning.
At that time I realised that in order to do what I value most - learning and solving problems - at its highest level, I would need to get back to the basics and focus more on learning than on doing, at least for a while.
So, now I'm back in uni, finishing my MEng and working part-time in a startup that I love.
Everything that I learned during my time working actually helped me a lot in my studies, to the point that I'm gonna graduate around the time that I would have graduated if I didn't drop out.
I also consider doing a PhD once I graduate.
Currently working as a SWE at HeyFIRE
Also trying to graduate NTUA.
On my way to break into the AI research / MLOps field.
Get in touch:
me@yiannisha.dev
I am obsessed with writing software, solving problems, building projects
and I will probably never stop doing it (and trying to get better at it) until I die.I FOUND MY CRAFT AND MY CRAFT HAS FOUND ME
MY ONLY GOAL IN LIFE SHALL BE TO MASTER IT
and I will probably never stop doing it (and trying to get better at it) until I die.I FOUND MY CRAFT AND MY CRAFT HAS FOUND ME
MY ONLY GOAL IN LIFE SHALL BE TO MASTER IT
Some projects I've worked on:
llm-tool
A simple Python module to automatically turn your functions into definitions that can be used for LLM tool calling. Built with Rust for blazing fast string parsing. Rust, Python
zspotify - Top Contributor
Spotify song downloader without injecting into the windows client Reached #1 trending on GitHub for a week Development stopped after cease and desist letter from Spotify Python, Ffmpeg
Amazon-Scraper
A command line program used for scraping data from different categories in Amazon's Movers & Shakers. Python
Startups I cofounded:
NAS-AI - CTO Feb 2024 - Jul 2024
Developed and launched a system combining hardware and software for real-time video streaming and AI object recognition, improving factory production efficiency.
-> Managed to hit very low latency with minimal hardware and infrastructure requirements (just a single CPU VPS on Digital Ocean).
-> Improved efficiency by eliminating the need for constant human observation. Used Python for core algorithms, C++ for critical hardware tasks, and Rust for secure back-end services, ensuring the software was robust and scalable. Used Docker for streamlined and automatic software deployments, enhancing system reliability and simplifying updates.
-> Managed to deploy updates, live, on production with no downtime. Custom computer vision bottle detection algorithm developed for a client's conveyor belt. The algorithm was parametricized and optimized using bayesian optimization for each use case.
Custom video streaming service
Python (OpenCV), Rust, ZQM, MQTT, C++, Docker, Raspberry Pi
-> Managed to hit very low latency with minimal hardware and infrastructure requirements (just a single CPU VPS on Digital Ocean).
-> Improved efficiency by eliminating the need for constant human observation.
-> Managed to deploy updates, live, on production with no downtime.
Custom video streaming service
Python (OpenCV), Rust, ZQM, MQTT, C++, Docker, Raspberry Pi
RobinStore - CTO Mar 2023 - Sep 2023
Set up and maintain cloud-based infrastructure for automated data scraping, cleaning and processing.
The scraping service was capable of scraping and processing all products from every single supermarket eshop in Greece at the time - every day. Implemented a product similarity service using Image Embeddings and a Vector Database. Developed a full-stack mobile app that users could use to find the cheapest option for each product they wanted to buy. Won the MVP award at VAIM 2023.
Eventually another company showed interest in acquiring the web scraping algo.
Web Scraping (JS & Python), Weaviate (Vector DB), Flutter
The scraping service was capable of scraping and processing all products from every single supermarket eshop in Greece at the time - every day.
Eventually another company showed interest in acquiring the web scraping algo.
Web Scraping (JS & Python), Weaviate (Vector DB), Flutter
Places I worked at:
HeyFIRE - Software Engineer Feb 2024 - Present
Manage the development of an AI Agent used by our users to automate tasks and analyse financial data. Build and maintain a client-side caching service which reduced the load time of the app. Develop and deploy features used by 100s users daily, including an onboarding achievement service. JS, Flask, Python, LLMs
IQ3Solar - Software Engineer Oct 2023 - Feb 2024
Set up and maintain infrastructure for the automated collection, processing and saving of the PV data. Integrate Machine Learning algorithms into the production, including automated training and live visualization. Build a full-stack platform for tracking, monitoring and analysing PV-related data and insights. Scikit, XGBoost, Python, NextJS, R
Wattz - Full-Stack and Blockchain Engineering Jul 2023 - Feb 2024
Maintain and Develop full-stack web decentralised applications. Test, debug, maintain and develop smart contracts written in Solidity deployed on the Polygon network. Develop processes for managing complex payment and transaction systems, combining both crypto and fiat transactions for users and the marketplace, ensuring a smooth and secure exchange of assets. NextJS, Bun, Hardhat, Solidity
AnvetoMarketers - Full-Stack Engineering Internship Oct 2022 - Jan 2023
Build full-stack, reusable plugins used by other developers in the company. Build a calendar plugin synchronized with the site admin's Google Calendar and allows site visitors to book meetings that are then automatically imported into the admin's Google Calendar. Configurable, scalable user registration forms paired with an admin page that allows each admin (based on their role) to view and edit user data. VueJS, PHP