Karim’s Thoughts

I’ve had the pleasure to work in many industries from an early age. My first job was at Mach II Maintenance at JFK International Airport. I performed line maintenance of some of the most icon aircraft to ever fly including the Boeing 747 and Airbus A380. Even now, when I think about it, I can hear the deafening tone and ground shaking startup of a GE90 engine. That career pushed me into engineering because I asked too many questions.

It turns out my inquisitive nature would lead me to my next 5 jobs at GE Aviation, Boeing, Microsoft, and Salesforce while I pursed my BS and MS in Electrical & Computer Engineering at Carnegie Mellon University. It wasn’t all work at CMU. I was one of the original founders of the university’s Mixed Martial Arts club. I also coached many student’s on career advice offering uncommon viewpoints like viewing an engineering careers just like a professional athlete’s career (do you want your jersey number to be retired or do you want to maximize career earnings?) and suggesting to apply to jobs under “white” names based on my personal data and Kang et. al. Chances are, between the satellites that I worked on, Xbox controllers I built, cloud services I’ve designed, and students I’ve coached into Big Tech SWE jobs, I’ve touched your life in some way, shape, or form.

Moving into Big Tech

At this point in my life, I counted 2.5 years of work experience but many recruiters disagreed. Apparently signing off on a couple hundred of lives daily and validating the Xbox Elite Controller for Phil Spencer to present at E3 in 2016 isn’t “relevant work experience”.

My determination to NOT start an entry level software engineer job led me to Salesforce. I also at the time wanted to get an industry research position in Machine Learning but there were several problems:

• I’d never formally taken a course on Machine Learning
• I had published only one journal paper in Computer Vision
• I didn’t have a PhD (which was more or less required at that time)
• I needed to bail myself out of student debt

So I took Mark Cuban’s advice to heart and I found the thing that I was effortlessly good at and spent a lot of time doing it. I built scalable, fault tolerance, high performance, low latency distributed systems. And I did it again and again and again until I literally built the cloud.

• Salesforce Application Model: This is where I started and I wrote a ton of code. Kubernetes was still upcoming and coming at this time and somewhere along the way had a breaking change. I did the upgrade for a team that hosted managed k8s clusters rewriting over 17,000 lines of code becoming the 3rd highest contributor on the team in my short 9 months there.
• Journey Builder: Salesforce Marketing Cloud was a behemoth acquired through ExactTarget. I tuned the monstrous processing engine that helped many companies achieve record profits for the 2018 holiday retail season.

The Sign-On User Experience for the Cloud

This is one that I’m particularly proud of because it led me to my promotion to Principal at only 27. I was one of the youngest to achieve the feat at Oracle. Here’s why:

Side note: When I joined, OCI was still very young and fun. There were no guardrails in place and it was very much like the stories I’d here of very early Facebook. You built fast and you broke things. This is the best advice I can give young engineers looking for opportunities. Look for places where the pace is just so fast, there are no formal processes in place, and you won’t get reprimanded for breaking things. In less than a month of when I joined, I caused a full regional outage because I pressed the wrong button. In hindsight, it was a learning experience for me and my manager because no one exactly knew what the button did!

Eventually, I had complained enough that someone had enough and said put your money where your mouth is. I was given my first team to lead and tasked with rearchitecting a sign-in service that at this point supported over 30 live production regions aka I was no longer aloud to break things.

And I did it all while improving OCI’s security posture and never causing a single outage for a single customer, in any region.

I did plenty more work for Oracle but you can interview me about that

Following my passion

I never lost sight of my research goals; I just had to take a small detour. All while working, I stayed in touch with an old friend at CMU (thank you André) and published some more work but in Formal Methods. I won the GEM Fellowship and got accepted to every Top-10 EE/CS/ECE/EECS PhD program in the country EXCEPT Stanford and UW.

I still ask myself if choosing Berkeley > MIT was the right decision but I’ll find out in a few years. I know for sure, that I greatly dislike the gym resources at Cal after being spoiled at CMU.

I enrolled at Cal, finally got around to take my first Machine Learning course (I jumped straight for 282 Deep Learning). I also was introduced to Reinforcement Learning by the infamous Sergey Levine. These days, I’m heavily involved in Generative AI (I love Diffusion Models but I do most of my work with LLMs) and RL. You can follow my Semantic Scholar profile for all my publications but I will occasionally share a post about them here.

Post graduation, I plan to get back into doing startups (as of 2023 I’ve launched 2 and we didn’t even get to talk about them!) and continue teaching and giving unorthodox advice as a university professor. Until then, I will continue to fill up my passport, learn more languages, work interesting jobs, and enjoy awesome life experiences.