Second Summary of my Journals: The Mountains We Climb

20 min readDec 29, 2021

Disclaimer: This is not a year in review. This is a weekly-monthly summary of stuff that had happened to me since the last time I published a similar article, as penned down in my journals.

Sequel to my last mountain article detailing my struggles, a few people have asked me to write an update to the article. I kept postponing the writing until December where it coincidentally looks eligible to be a “year-in-review” article. Most year-in-review articles I have seen are pretty great stuff. I myself had a very good year. But sweeping all the struggles and pains under the carpet while listing out all the “wins” is not what I am doing here. I’d be stating things as they happened to me, with very few filters. I feel it’d make a better read that way, and hopefully, help you relate to my story better.

October-November 2020

Did more of AI community work. Took up my time, energy, and soul. The highlight was when AI Abeokuta won the “AI City of the Year” award, and I was the 1st runner-up for the Mr. Algorithm award which is an award for “the most outstanding male starlet in Nigeria’s AI space who not only strive to be the best but also invests in other through knowledge sharing and community engagements”.

December 2020

I got feedback from the HR manager at Microsoft Africa concerning the SWE intern role that I applied for — I’d be having my first interview in December. I did a coding assessment, completely finished 2 out of the 3 questions. Afterward, I had a combined “getting-to-know-you” and a Data Structures & Algorithm interview but I didn’t make the cut to the final interview (got the mail in mid-January). I think the main takeaway from this was that my resume had gotten strong enough to land interviews.
I started taking my fitness and energy seriously by working out consistently. I was mostly concerned about getting my energy levels up, having nights of good sleep, and trying to avoid avoidable health issues when I advance in age.

January 2021

It's more of a low-key period for me. Reading Atomic Habits and trying to inculcate its teachings in my everyday life. I decided to make exercising a habit too-I decided not to miss two exercise days in a row (worked for me until that October stress).
Working on an end-to-end machine learning application I called ApartRent. My web scraping and data wrangling skills took a big upgrade during this period.
I started contributing actively to open-source via Data Umbrella’s outreaches with scikit-learn. I had a mind to participate in GSOC later in the year.
I started taking a graduate class with the University of Nebraska-Lincoln on computational game theory, courtesy of DSN. Stopped halfway in March due to other pressing pursuits.

February 2021

I was still contracting with Scitylana Ng. The money wasn’t too great but I was working as the sole executor of projects and that was very nice and challenging — I stopped in March though as I got disillusioned with the work conditions. I just wrapped up an E2E ML application for one of our clients, a lotto forecasting company and it was pleasing. I even had the company’s CTO write me a recommendation on LinkedIn.
I did my interview with iQube Labs, after submitting my assessment test. I had insider information that I’d get an offer soon but I didn’t. They sent me a rejection mail after I followed up severally… One of the painful moments of 2021 — I wanted it so badly!
I started my GSOC preparations after iQube’s heartbreak. I started contributing to CHAOSS and participating in the meetings.

March 2021

One of the funny moments of the year; In my Software Engineering class in school, we were grouped into several groups and asked to design a product for our CAT. My group selected a project but they were too lazy to work on it so they suggested hiring an external UI/UX designer to do it. I was mortified by the idea — we are CS students, for crying out loud! So after getting a quotation from a professional designer and my group members paid up, I learned enough product design and Figma to get the job done — none of my group members the wiser. Turned out that our product idea was implemented by 3 other groups so we decided to work on a new product within the span of 48 hours. The deadline was too short so I wholeheartedly hired a professional designer this time around!
Habeeb Shopeju introduced me to Deep Discovery’s CTO for their machine learning intern role. We sort of hit it off over the emails and we scheduled our first meeting. I couldn’t make the first meeting due to my exams in March and we scheduled another meeting which he wasn’t able to make due to the Pfizer vaccine he took. Our correspondence sort of lost steam and my attempts to revive that later in April didn’t work.
I had my first meeting with the mentor assigned to me by DSN, Dr. Babatunde Olorisade via its mentorship program. He is a Senior Lecturer of Data Science at Cardiff Metropolitan University. I told him about everything I was working towards, and my aspirations, which are; open-source development, machine learning engineering, machine learning research, and PhD. Under his guidance, I started working on a roadmap towards achieving them.
I was still actively contributing to CHAOSS. Venu Veddy, a GSOC mentor in CHAOSS, was extremely resourceful and helpful throughout the process. My understanding of Git’s capabilities as well as the project’s internals grew rapidly during this period.
I was one of the top contributors in CHAOSS. Unfortunately, one of the remaining top contributors was extremely interested in the same project I was targeting. So we’re both trying hard to outdo each other.

April 2021

Typical email correspondence with my mentor, Dr. Babatunde. We were having bi-monthly meetings/email updates.

I submitted my project proposal for GSOC. I got a rejection mail later in May. I had a follow-up chat with Venu and he told me that it was a collective decision and some of the mentors felt that the project was too complex to be completed in 2 months so they scrapped the project. This was the most painful moment of 2021 because I put in lots of work and a huge part of my next 6 months was built around GSOC happening for me. I learned a lot, regardless.
I reshuffled my roadmap after GSOC failure so the focus was just machine learning engineering and research. For the engineering part, I decided that I’d participate in the DeepQuest Sign-to-Speech challenge. For the research part, I started taking foundational courses in the Deep Learning and TensorFlow Advanced Techniques Specializations.
I started applying to companies for my SIWES internships. Since Microsoft didn’t work out, my mind was mostly fixed on DSN Research Team, as I believed it’d aid my research journey and PhD goals. I had an interview with the team and I was told they’d prefer a graduate for the role, the fact that I already had graduate-level experience and I had my next 6 months free of school commitment didn’t matter to the team. I was more annoyed than upset — I had been a super-active community leader with DSN since 2019 and I had always thought that I had a “fallback” option there.
InstaDeep and DeepQuest were the only other 2 Nigerian companies that I was very interested in, due to their research tendencies. Winning DeepQuest’s AI challenge would make getting into DeepQuest straightforward for me. For InstaDeep, I used LinkedIn massively. I cold-mailed everybody from the CEO in its headquarters in London, Karim Beguir, to the company’s research engineers in Nigeria. Everybody that responded positively referred me to InstaDeep Nigeria’s Team Lead. I told him about my aspirations and shared my resume while following up regularly. Then one day, he gave me a phone call and told me that they’d prefer graduates for the internship roles because they’re looking to expand and retain them permanently — sounds fair! Then he suggested that I join the next training cohort at AI6 Lagos. That was when I became so sure that he didn’t even go through my resume. If he had, he’d have noticed that I myself have organized and facilitated ML training programs at the level of AI6’s. I just smiled!
At this point, I was getting pretty tired. I didn’t really understand why everybody was associating being an undergrad with “a noob that you’d have to spoonfed and show the ropes every time”. I had 4 years between high school and college, so by the time I started my first year, I was already decisive with and taking steps toward my career. But I guess that wasn’t visible in my applications.
It’s an application galore as I applied more for my SIWES placement— Verraki Africa, Voyance, TeamApt, and I got a connect into CapeAI, a South-African company. Verraki requested my CV and didn’t get back. Voyance, the same. TeamApt’s HR manager told me that they are not taking interns at that moment. I did 2 interviews and 2 technical assessments with CapeAI and I wasn't given feedback. By this time, I decided that I’d just do my SIWES at home.

May 2021

This month was extremely low-key — I was working on my skills, mostly. I was neck-deep into the Deep Learning Specialization on Coursera, to build my foundational knowledge for research. I took TensorFlow Advanced Techniques Specialization too. And I started working on my DeepQuest’s challenge on sign-to-speech.
I was also organizing the 2nd cohort of our mentorship program at AI+ FUNAAB around this time. Since a good number of the mentors were not physically present in school, we made the whole thing virtual, modeled after She Code Africa’s mentorship program and AI6 cohort.

June 2021

I finished my sign-to-speech project and submitted it to DeepQuest via a well-documented GitHub repo — my open-source knowledge is not going to waste. :)
I realized that working and learning at home won’t cut it for my SIWES as I have to make it official. I decided to offer my services at workspace Hub close to my location, ATC, in exchange for offering me the hub as my SIWES placement.
I finally got a data science internship offer with a Connecticut-based firm, SeqHub Analytics, through my friend, Victor Ojewale. I started working with them in mid-June.
As providence would have it, my first student at ATC, Omolade, who registered for graduate training in ML at ATC, was an incoming PhD student at a US university. She graduated from my school — she was in 400level while I was in 100level but I didn’t know her from school. She demystified the whole process for me including the 2 exams I have to take, the fundings options, and how to look for schools and advisors. She explained how her process was and that was when I realized how achievable the goal is (knowing somehow who did it from a similar background as mine certainly did help) and I planned on taking actionable steps towards it so I’d be in the position to apply immediately after my BSc (next year). I had been contemplating abandoning NYSC after college. This was when I finally made the decision to skip NYSC.

July 2021

I drafted out my comprehensive roadmap for my PhD plans. The plans included participating in Kaggle competitions and doing a FAANG internship or AI residency early next year before I start applying in fall 2022. I started looking at “safe” schools in the US too while carrying my mentor along.
I started preparing for the GRE exams an hour a day. Started taking mock tests too and my scores were just a little above average. Not shabby!
My Sign-to-Speech won the DeepQuest AI challenge. I screamed ehn! My 2021 wasn’t going as planned at all — too many rejection emails as at this point and I needed a win so badly.

My first big news of 2021! Didn’t come until July’s ending. It was crazy from here on though :)

Opeyemi hit my DM to congratulate me and I asked if I could scale it up to become a research project. She affirmed that and we scheduled a call. She suggested that I could bring Dr. Babatunde onboard.

I informed my mentor about Opeyemi’s suggestion — He asked me to do a literature review to see the space the work could fill. Opeyemi and I did and decided on focusing on Nigeria or Africa, and he was convinced enough to team up! :)

August 2021

We started having a flurry of meetings on project scoping. This period was a little bit discomfiting as I was getting asked critical questions, half of which I didn’t have answers to. :(
Around the same time, Sebastian Ruder’s newsletter for July’s ending came in and he mentioned research communities doing awesome work. He mentioned Masakhane, Eleuther AI, and ML Collective. I was already familiar with Masakhane so I checked up Eleuther AI and MLC. MLC was going to have a research jam in two days' time and even though I didn’t clearly understand the sort of project I could present, I submitted my sign-to-speech project for presentation. It was a 5 minutes presentation and 8 people/teams presented. You could say that mine was the least research-centric amongst all the presentations! :)

Rosanne Liu, the executive director of MLC, reached out and told me that MLC could support my research paper quest (I mentioned in the talk that I’d be considering scaling up the project into a research project).

ML Collective is a really fantastic place to be in! I ditched my plans of doing research via Kaggle competitions and focused on MLC full-time. I started joining the weekend NLP and CV reading groups. I also joined the NLP study group taking CMU 11–747 course and meeting to discuss it on Tuesdays.
Rosanne started the CSS program for ICLR, aimed to bring underrepresented researchers together to submit to ICLR. I joined up and I got a few potential collaborations. Dr. Giselle, Swapnanil, and Nuredin Ali. I later dropped all as my research project started claiming more of my time. I also met Nils, a German PhD researcher at DFKI. He was looking for a research assistant so I signified interest. He said his lab is government-funded so he won’t be able to pay me since I’m not based in Germany, but I decided to take the role up in a voluntary position, regardless.
MLC reading groups use the role-playing format for reviewing papers, so I volunteered as an archaeologist on one of the reading sessions for computer vision. Quite a milestone!
Dr. Abiodun Ogundeyi (medical doctor kind) came to visit me at my IT placement. He was leaving for his professional Master's in AI in the UK so he came to say goodbye. I updated him about my progress with my PhD plans (he was one of my earliest gingers in the whole grad school thingy since 2020) and he suggested that I could ditch TOEFL by writing to the school. So the only exam I needed to write was GRE.
Opeyemi linked me up with the education director at Afrisnet, to support my grad school applications. I was part of the last batch of mentees that Dr. Schwab took on later in September. Lucky me!
Finally, I was announced as the GDSC Lead in my school. I started making plans to assemble my core team.

September 2021

Rectangle AI decided to give me an offer letter as an AI Engineer. We were to finalize discussions on resumption in October.
I resumed DeepQuest’s internship and it got terminated after a week. Apparently, I was still in school and doing a part-time, voluntary research assistantship which I didn’t inform them beforehand so they decided that I won’t be able to give my all to the company. I was relieved, actually. I wasn’t even upset at all. I guess it’s because I just got a full-time position so I felt an internship isn’t the biggest thing to happen to me.
I created my first sign language dataset with Amanda, a young TV sign broadcaster at OGBC. Arlee, a hub user at ATC saw me kneeling down and raising up fingers :). He asked me to explain and later suggested that he could connect me with a professional signer, Amanda. Amanda was so sweet and patient. We created about 3000 images of the dataset.
I performed the first iteration of experiments and submitted the first paper draft to AAAI-22’s Student abstract program.
I created the second batch of the dataset with the students and teachers of the special education department of Saint Peters College, Olomore. Anjy, my friend, connected me with a teacher at the school who connected me to the HOD for the special education department. The HOD fell in love with my prototype and scheduled 4 class sessions after their break time for the dataset creation. Those hearing-impaired students were so energetic and they smile a lot!
I presented my project update at the next MLC’s research jam. Lots fo feedback and suggestions, and since performing the research experiment on Colab was a pain in the ass, MLC gave me computational support (GCP) to push forward my experiments. Yay!
Had my first call with Dr. Schwab. He told me that my application profile is quite good, due to my technical experience. And he gave me enough reasons why my profile is good enough not to bother with writing GRE — a relief! I was getting bored reading for the exam. After weeks of several iterations, I finally created a badass academic CV that pleased Dr. Schwab.
Thao Nguyen, a PhD student at UW, started mentoring me and my applications. She was the first person that made me realize how unique my profile is — a brilliant student from some unknown school in Nigeria/Africa, stepping out of his research-lacking environment to hone his research development abroad while also leading his local tech communities with aplomb. She graduated from Stanford, did an AI residency at Google Brain, then got into UW. Yet she labeled my journey as “more interesting” than hers. Sounded like a fairy tale, but it was a huge confidence booster for me! I started pursuing the really ambitious schools and dropped my “safe” schools one after the other.

October 2021

Finally wrapped up my paper and submitted it at the NeurIPS workshop on ML4D. Took my sweat and blood, the submission! Nayan Saxena had to step in at some point to help with the draft editing and LaTeX conversion. I wouldn’t have submitted without him.
I started famzing Colin Raffel, my first choice advisor for grad schools. He told me he won’t be taking in any new students. I was so captivated by his work on T5 so this was quite disappointing. Started reaching out to professors en masse, sending my CV and research profile.
I applied for the ML Developer Advocate role at HuggingFace. Famzed with the Advocacy Lead via MLC. I did 2 interviews and an assessment. In the end, I was rejected! I had a feedback call and the reason I was dropped was explained to me — I wasn’t technically strong enough. Fair enough since I was competing with the likes of Rising Odegua, who is a Github Star and a GDE in ML with tonnes of open-source track records, for the position. I was pained because it had all the roles that I hold close to my heart in one place — ML research, ML engineering, open-source development, community, and advocacy — but making it to the final stage was a testament to how far I’d come.
My paper was accepted into the NeurIPS workshop on ML4D! Yay! All four reviewers gave strongly positive reviews on the paper. All the hard work paid off! Rosanne and Jason blew the news beyond proportion and I became twitter-famous overnight — a young Nigerian undergrad cum wonderboy from an underrepresented community who didn’t let anything stop him to acquire research experience and publish his first work in a workshop at the most prestigious ML conference in the world. Even Masakhane invited me to come and speak.
I also submitted the work as a poster at the DSN Bootcamp 2021. Even though my table was flooded with audience throughout (and I didn’t even mention the NeurIPS acceptance at all), I got the second-best poster award. The OAU folks have pulled their strings once again! :(
Since my paper was a success, I turned my attention to grad schools and started writing my first draft of my SOP. That first document was a mess, lol. And finally resumed back in school!

Paper acceptance at the NeurIPS workshop on ML4D! Whoop whoop!!!

November 2021

The 2nd-Best Poster Award at the DSN AI Bootcamp

I finished writing the camera-ready version of the paper and published it on arXiv! Turned out that I needed an endorsement from an actively publishing researcher because I could publish my paper so Rosanne had Mitchell endorse me. Mitchell schools at UW and followed me on Twitter so I followed up, had a call with him, and he helped suggest best-fit potential advisors at UW and even introduced me to 2 of the advisors. The guy is so influential!
Lelia Marie Hampton, a PhD student from MIT, started helping me with my application. Just like Thao, we became really good friends in the process. Connected me with Dr. Bau of Khoury college and also connected me with the only 2 Nigerian guys in MIT’s EECS PhD program, Julius and Adedayo. Both were extremely helpful and encouraged me a great deal. Adedayo even paid 2 of my application fees.
Rosanne and Jason agreed enthusiastically to write me recommendation letters, yay! My mentor from DSN, Dr. Babatunde Olorisade was also the first choice so no lecturer from my home school would be writing me letters. I needed only 3. But Stanford required a home school professor so I had my supervisor write one for me later. I’m very sure that the letter he wrote is a “did-well-in-class” letter :(
Had additional calls with Julia Belyakova of Khoury college, Nader Akoury of UMass, and Jack Kolb of GaTech. They were all super-helpful and helped refine my SOP and/or helped in identifying potential advisor matches.
Attended Young Professional Bootcamp to wind down after a few super-hectic months. We stayed at our lodge for 3 days — I really did enjoy myself and I got to know Vivian, my traveling companion, on a personal level.
I resigned from my work at Rectangle AI. My resumption at school and the chaotic schedule made me so unstable that I couldn’t deliver on my assigned tasks on time and I had to think thoroughly about what I wanted next. Moyosore was extremely helpful. Grad schools application was the highest on my priority list followed by research and school, so I resigned.
Attended 2 DevFests and several wedding ceremonies. I was really starting to wind down!

December 2021

After lots of painful iterations with Dr. Schwab, I finally had the perfect SOP. Started submitting my applications while updating my recommenders on my submission progress. Created a spreadsheet for them to track the submissions.
Met Dr. Bau and Prof. Katharina Kann. The calls were so interesting and reassuring — I realized I was being nervy for no reason! Katharina even connected me with one of her current students so I could have a one-on-one chat on work-life, et cetera. And that one, Sagi, discussed my cities options with me and had me daydreaming about cool office spaces, mountains, hiking, snow, pale sun, beach, more mountains, snowboarding, cycling, et cetera. Half of my mind has already left Nigeria lol.
Wrapped up applications and I started winding down for real. Movies, booze, flings, and outings. Ending the year in flex mood.

Biggest Takeaways

My communities, my life: I realized how integral communities are deeply woven into my life and endeavors. ML Collective was my most impactful community of 2021 and I still wonder about how an organization could be so beautiful. And I also wondered why Thao could have been so selfless toward me and went beyond PAMS’s duties, or why Lelia put in massive efforts beyond GAAP’s responsibilities to assist me. But then, I have always worked hard in helping people in my local communities too. Most of my mentees from She Code Africa, DSN Abeokuta, the general DSN community, AI+ FUNAAB, and our local Google Student Developers club are my closest people now. A good number of them named me as “the person that influenced you the most in 2021”. And I remember one of our mentees at AI+ asking me if we were being paid by our sponsor organizations to mentor ML enthusiasts, due to the way we carry our mentees’ lives and dedication on our heads. :)
Consistency and properly structured, flexible goals can do wonders! What I was doing by the end of the year was about 70% different from what I had thought 2021 would look like.
Compound interest is real. Most of my achievements were built on each other. And I didn’t achieve any visible win until the ending of July! From that point onward, it was a frenzy period for me!
Success has many friends! It is quite easier to receive bits of help or make valuable friends when you are winning or are showing serious potentials.
Gratitude journaling still working for me! 3–5 mornings in each week, I’d write about the 3 (or more) things that I am grateful for, the one thing that got me upset, the one thing that I am excited about doing that day, the one thing I’d do that’d propel me in the direction of my dreams, who I could serve that day, what my purpose is, and if my goals are still connected to my purpose. This leads to the next point;
Many different intellectuals and experts have argued and exchanged ideas on what they perceive “purpose” to be. For me, my purpose is my obsession and what is consuming me for the time being (usually months, based on 2021). It is what most of my goals during that period are oriented towards. In some parts of 2021, my purpose was to become an employable machine learning engineer with a firm record of E2E applications and open-source development. In most of the latter part of the year, my purpose has slowly transitioned into being a machine learning researcher that can get PhD admits into top CS programs in the US. Since that purpose was achieved, my purpose is experiencing another change currently.

Regrets

I have a growth mindset which means that I rarely beat myself up over my wrong decisions. I’d just learn from them and move on. The only thing I wish I’d acquired in 2021 (while not acquiring it) is a stable and high-paying job by the end of the year, and I don’t have that yet.

An afterthought… Somebody asked what I’d have done differently to get into GSOC and I mentioned that I’d have selected a simpler project. The project I selected was the toughest project on the platform — an expert system for a rule-based recommender system in Grimoirelab’s SortingHat platform (a good blend of Django, Vue, and GraphQL). I wanted to flex my SWE muscles so I picked it. Whereas there was a “medium-sized” ML project that was to improve Augur’s ML methods — right in my safe zone — but I ignored it. Only 1 person submitted a proposal for it and his proposal was quite half-baked enough to make me cringe while going through it. But come to think of it, if I had gotten selected for GSOC, would I have had enough time to work on my research project and position myself in a sweet spot for PhD? Your guess is probably the same as mine.

In conclusion…

2021 was a pretty good year for me. I didn’t xx my earnings in the year like most of the professionals in the tech community writing their year-in-review articles said they did (I still don’t have an earning lol), but I couldn’t have imagined how huge and unbelievable the year would have been for me beforehand. I discovered a new height to my consistency level. I got my name in the global space and upped my game seriously. I made lots of new friends and important connections. I lost very few friends too. And I tried to help my people along even as I climbed higher up my development ladder. I experienced very strong feelings for a female even though we had to cut ties. And I had serious fun in the year at different levels. And hopefully, 2022 promises to be pretty unbelievable too.