I'm looking for a remote programming job, likely backend or working with data pipelines or doing data science. I'm highly skilled with Python. Also, I have experience with data mining / web scraping. I'm familiar with LLM/AI APIs and how to both train and use them, and I will say they are extremely efficient for structured data extraction from unstructured sources like websites and PDFs. My main tools for building AI models are PyTorch and LightGBM.
I don't want a full-time job so much as part-time or flexible hours. I can get a great amount of work done in little time, so I'll probably prefer fixed price contracts to hourly pay. I have over 150 jobs via Upwork and have had success with jobs involving maintaining regular data ingestion. Automation scripts are another task I really enjoy.
I can generally work with almost any technology given some time to learn it. I've done static websites, front-end javascript, SQL, MongoDB... anything involving using existing APIs is generally very easy to set up too.
2005-2011: High school (graduated) and college (quit 2 years in). Was top mathematical talent and participated in the USA Math Olympia (USAMO) in high school.
2011-2018: Ran an online store arbitrage operation where I would buy things at a low price, using alert systems and eBay & Amazon APIs, then resell them online at market price for a decent profit. Quit because I got sick of being surrounded by the merchandise. The tools I made here I mostly used PHP for.
2016-2018: Studied machine learning models, how they work and how to build them. Did several Kaggle competitions and several online classes. Learned PyTorch and implemented some original models/layers from research papers. Also learned Python in the process.
I also passed actuarial exams P and FM around this timeframe.
2019-present: Worked as a freelancer on Upwork, grossing over $100,000 of income across over 150 jobs. Most common jobs were web scraping and Python related jobs, with just a few machine learning jobs in the mix.
One of my most complex jobs involved ingesting information from an aggregate news feed, running a classifier on the article titles to get just the relevant ones, and using OpenAI's API to run unstructured data extraction on the articles.
Although those 3 time periods represent my paid work experience, I've done a ton of unpaid / fun projects too. Some memorable examples below.
1. Photograph editing script to get good looking white background photos out of raw shots. (Used for the online store.)
2. Many video and image editing scripts using ImageMagick and FFmpeg. One example of a fun one is a script that takes a video, splits it into a specified number of different segments, say 6 x 4 = 24 segments, and then puts a fraction of the video in each position on the screen. It looks like a security camera monitoring setup. (This would be useful to skim a longer video in a shorter amount of time.)
3. Made my own Chrome extensions for various little functions, like a shortcut to pin a tab and a script that keeps pinned tabs from being closed by the close tab shortcut.
4. Have made many static websites, including RSS feeds of live scraped content. This came in handy for some of the jobs, too. One such newsfeed that slightly came in handy outside of work was a feed of local houses for rent, at a time when I was looking to rent one.
5. Built a product image classifier for the store (never used).
6. Built an image dithering program and set it up with a thermal receipt printer so that the printer would print high fidelity images. (Had to play around a lot with sharpness/contrast to get this just right.)
7. I like books, so I made some scripts to be able to print PDF books in 2-per-page or 4-per-page size on a standard 8.5" x 11" sheet. By the end, I was able to cut sheets, punch holes, and align the stacks so that the pages were in the correct order for the book.
I'll also add that printing on double-sided vinyl paper is a great way to have waterproof pocket-size copies of a favorite book.
8. I've also done a lot of work with MIDIs and generated sheet music (via LilyPond).
9. I recently tried a search engine by taking downloaded index page HTMLs, generating keywords/tags via AI, and making a tag cloud of the keywords. All this proved is that it's hard to make good search engines. But with the right kind of data and tags, this could be extremely useful.
10. I've also made a lot of userscripts (Javascript scripts that tweak browser functionality) and AutoHotKey scripts. I've gotten real work done with AutoHotKey. But some of my favorites are scripts to add to browsers are navigation keys, e.g. pressing left/right will go to the previous/next page of a forum or news article list.
1. Bulk extracting structured data from web articles or PDFs using LLM APIs.
2. Creating an RSS + HTML news feed of anything that's worth monitoring. This includes real estate.
3. Building machine learning classifiers, even complex/original models. For instance, a random forest classifier on a bag-of-words of good vs. bad texts. This was how I did the news article classifier.
4. I'm extremely proficient with web scraping, which is a rare skill to be good at. I can do SEC/Edgar data and real estate data very well.
5. I can do LinkedIn lead generation very well. Basically, if you had a target segment you wanted to do an email or call campaign on, there are ways to pull email and phone contacts using their LinkedIn profiles.
5b. I also have some other techniques for getting tons of accurate business addresses and phones.
6. I would say my best work, though, is in setting up and maintaining regular/recurring processes. I had one employer who said he wanted to continue working with me, but after I showed him my system and he figured out how to use it, that was the end of our business. I hope to leave many clients as happy as this one.