2019 in review

by Kyriakos Chatzidimitriou | Mar 23, 2020 18:26 | life

2019review

Finally! I had this post in draft mode for about three months, but better late than never. This post also signifies the start of blogging in 2020!

2019 is (well) over and through this post I will try to summarize important aspects of my life that can be quantified and reflect on what went well, what didn't and what I've learned this year the James' Clear way. As in my 2018 review, the first part is pretty quantitative, while the second is more qualitative.

In 2019, and after failing a couple of tenure applications, I switched from the Academia to the Industry and also left Cyclopt, the spin-off company I co-founded. In fact I switched two positions in the Industry, back-to-back, because the first company I signed up for (a mature start-up) got out of funding just one month after I started working. In my job-hunting adventure that started early in the summer of 2019, to get a couple of offers, I applied to around 80 positions. I will leave the details and the stats for another post.

In general, I don't consider that 2019 was a good or lucky year for me: My father passed away and I had to look for a job, one month after starting on one. But as Nietzsche said: "What does not kill me makes me stronger".

Quantitative

Books

Even though I started a lot of books (both non-fiction and tech ones), I only finished:

  • Atomic Habits by James Clear. Has all the theory, the systems and the processes you need to start new (better) habits, but as with many other personal improvement guides the problem I have is in execution. I will definetelly re-read it, since I read it during my summer vacations and this is not a good time start building new habits.

"Launches"

I launched a couple of web sites and apps this year either alone or with teams:

and some open source software:

  • gh-downloader for downloading files from GitHub given search criteria
  • character-position VS code extension for revealing the current character position
  • eslint-config a shareable config for web application development using node.js and react.

Talks

I gave two talks:

Courses

I taught two courses in the university:

  • Big Data Analysis to graduates
  • Software Engineering to under-graduates

Research Proposals

I submitted one proposal for research funding as a principal investigator to ELIDEK (early in 2020 I learned that it was unsuccessful).

Blog posts

I wrote 12 blog posts, much, much better than previous years:

  1. 2018 in review
  2. Making env vars more robust by making them more fragile
  3. On collinearity and feature selection
  4. Calculating the running average and variance of streaming data using redis
  5. Advices and strategies I learned from my first business attempt
  6. Simple rules for building robust machine learning models
  7. Launching the new kyrcha.info using Gatsby, Bulma, Contentful and Surge
  8. Sending graphql queries using http.Client in Go
  9. Fitting modified Gombertz and Baranyi equations for bacterial growth in R
  10. Generating plausible paper titles with Recurrent Neural Networks
  11. What is a (startup) mastermind group?
  12. Data Outlier Detection using the Chebyshev Theorem - Paper review and online adaptation

The website had 2,705 users visiting vs. 5,189 in 2018. I am not sure abuot the drop. Probably also had to do with the switch in technologies from Wordpress (with SEO plugins etc.) to GatsbyJS.

Publications

Published 4 papers out of 8 submissions (50%). The published papers were:

  • "npm-miner: An Infrastructure for Measuring the Quality of the npm Registry" in MSR 2018
  • "Predicting hyperparameters from meta-features in binary classification problems" in AutoML 2018
  • "A Natural Language Driven Approach for Automated Web API Development"_ in WS-REST 2018
  • and "Deep Reinforcement Learning for Doom using Unsupervised Auxiliary Tasks"_ in arxiv

Competitions

  • Really worked with Kaggle Humpback Whale Identification Competition and learned new stuff even though the approach did not generalize well. It was my first image recognition pipeline ever written so I've learned a lot.
  • Worked on the CodRep 2019 competition, 2nd place. The writeup of the compeittions can be found here.

Qualitative

Things that went well

  • Output in terms of blog posts, papers submitted, competitions participated, software produced and launched, job applications submitted, interviews conducted.

Things that didn't go well

  • Unfortunatelly we didn't get any VC funding in 2019 for Cyclopt after applying and following the processes of 3 VCs, so I decided to leave the start-up and focus on other aspects of my career.
  • A recurring theme, my weight and in general the fact that I didn't weight-lifted as much as I wanted (less than 2 times per week).

What I've learned

The revelation after re-reading "the subtle art of not giving a f*ck" that to be happy, solve problems you enjoy solving. Life is suffering. You will suffer. So at least for the problems you can pick, pick the ones you enjoy solving. I also created a slide for one of my talks for process:

General advice

If you want to do a career in Academia:

  1. Don't do work you would do in a software company. Take on projects that will have research outcomes and not just build applications.
  2. Follow the publish or perish rule.

I learned about Naval Ravikant and devoured a lot of content from him, especially the viral tweetstorm "How to Get Rich (without getting lucky)" and its related podcast.

I don't remember where I caught this but when you press submit to deploy an application you shouldn't prepare your suitcases just yet.

2020

No goals for 2020. I am setting up processes and systems, building atomic habits, shipping early and often.

Previous reviews

Photo from mohamed Hassan from Pixabay

Previous