From Agrarian to Data Scientist - A transition

In this blog post, I’ll share my journey of transitioning from a plant disease expert to a data scientist.

Back in 2018, I read an article about Freelancing in The New York Times. The author anticipated that people would start working from home in the coming few years than doing 9 to 5 jobs in offices. The author was correct.

Time flies and right after two years, I saw this coming true. After graduating in December 2019, I came back to Pakistan to serve in Academia, as per the Fulbright scholarship’s requirements. However, due to the Covid-19 pandemic, recruitments ceased; thus, I couldn’t find any opportunity to manifest my research and quantitative skills.

While surfing on LinkedIn one day, I found a post, from one of my connections, about Fiverr that I usually opened in curiosity. It was my first encounter with any of the freelancing websites. I skimmed through the site and found that many experts offered their gigs in many sections of life, ranging from graphic designing to content writing and video editing to career counseling. After a brief overview, I searched for services being offered in the field of data science, statistical data analysis and Geographical Information System (GIS). I was amazed that many experts were offering their gigs in these segments. After reviewing profiles of few good-rated data scientists, I learned that I hold some skills and expertise in this field that I can offer. Thus, I signed up on Fiverr made a basic profile that I kept polishing gradually.

On Fiverr, I started working as a data scientist and GIS & Remote Sensing expert. It took me two months to secure a small order, but it was worth winning it. After that, my profile came in search. I was contacted by different independent firms and businesses for some small-scale data science work. In the first two months of working on Fiverr, I finished around ten projects and fulfilled all of the requirements to become a level one seller, an improved version of the selling category.

Later I started exploring other Freelancing platforms and came to know about Upwork. After some reading and knowing about the platform, I signed up on Upwork too and successfully hunted some of the good projects that helped me build my profile. Note: I bid dozens of proposals to make my project winning likelihood high.

Freelancing helped me in learning things beyond my field of expertise.

In my Ph.D., I was actively involved in learning new tools like ArcGIS, R and Python. However, I applied these tools only to my own research data analysis. I was able to run various data visualizations and statistical analysis tests such as mixed models, regression models and some machine learning techniques but for my research data only. I never anticipated that I’ll ever use these tools outside my field. However, freelancing made me learn new skillsets in statistical data analysis and data science and expand my data knowledge beyond agriculture. In the early months, I learned many state-of-the-art data analysis packages in R, such as Tidymodels. Working as a freelance data scientist, I got projects from different walks of life such as hoteling business, engineering processes, stock market trends, Airbnb and health data, to name a few. Interestingly, few of these projects had data in a couple of gigabytes that also made me learn to handle such big data.

Also, the projects I worked on were not all about data analysis; some required building interactive dashboards, primarily related to COVID-19, in MS Power BI and ArcGIS. So these tools were also worth learning.

Besides statistical analysis, I also honed my skills in data cleaning and data wrangling. Because most projects had data sets that were rough and in a raw format and required substantial data cleaning before moving towards data analysis, these were the times when I thoroughly explored the functionality of the Tidyverse library of R, a famous data science R package.

Besides data cleaning and advanced statistical analysis methods, I also learned to make professional RMarkdown reports, interactive dashboards, building repositories on GitHub and publishing data and code on open-source framework (OSF).

Freelancing gave me potential growth.

As Miguel Cervantes once said that “Don’t Put All Your Eggs in One Basket”, now after few months of freelancing, I was ready to explore other baskets to put my eggs, meaning finding avenues to expand my client base and generate new income streams. This is where my reading habit helped out. Besides reading research articles, I also invested some time reading blogs and listening from other freelancers about other ventures that I could explore besides Fiverr and Upwork for data analysis projects. Resources like Medium, Quora and Youtube helped me significantly in this regard.

The first thing that I was revealed while looking for new avenues was online tutoring and the very first website I came to know was Wyzant. As usual, I explored the site briefly and signed up. Then, I started submitting proposals to students’ requests and also tried answering some of the questions posted by students to get my profile noticed. After few days, my profile appeared in search and a student contacted me about tutoring her some basic concepts in Epidemiology. I still remember that I had a two-hour session with this student and made my $60 ($30 per hour). After the first few sessions with some students, my profile was on the first page of tutors in Biostatistics, Epidemiology, and GIS subjects. Note: Tutoring is a bit hectic, but it generates more income in less time than data science projects, which can be gruesome.

My quest for exploring avenues does not end here. Recently, I come to know about Kolabtree, a freelancing site made for scientists and industry experts. Kolabtree is different from other sites as it is yet a smaller and actively developing platform that mostly hires PhDs and industry experts. Also, projects on Kolabtree offer more money than other sites, which are now being considered cheap labor sites. Once again, after acquiring some brief knowledge about this site, I signed up as an expert and started bidding for statistical data analysis projects. Luckily, I got few projects that added a significant sum of money to my earnings. This way, another of my income source came live.

The third significant advantage of freelancing I experienced was getting research collaboration offers.

Earlier this year, I joined Kaggle, a subsidiary of Google LLC, an online community of data scientists and machine learning practitioners. After that, I started contributing to different data science competitions. To showcase my work, I shared a couple of my notebooks on various social media platforms, and I believe this was the time when some of my contacts noticed my coding and data analysis skills and offered me collaborations that will lead to scientific publications. Now, I regularly receive offers from national and international research groups to join them as a data analyst and provide feedback on experimental design, statistical data analysis and visualizations. Usually, I lead the data analysis part in these scientific collaborations. To date, I am part of five different research collaborations, all from different countries, and hoping to be an author in some good quality publications.

Recently, I am offered a joint position as a data science fellow at The University of Exeter and The Alan Turing Institute that I’ll be joining later this year. All this is due to my efforts and hard work in learning these new in-demand skills.

This was a brief story of my transition from an agriculture expert to a data scientist. If you enjoyed it, please share it, and if you have any comments, please mention those below. Also, if you have a data analysis project for which you need help, email me at mohsin1570@gmail.com, and I’ll be happy to discuss it.