Aaron Fulkerson

Yay! I am retiring Friday, June 26, taking Microsoft's VRP (Voluntary Retirement Package) after a wild career as a political aide, computer salesman, gold miner, carpenter, waiter, and computer / data scientist. Thank for all your help since I went back to school (UNC-Chapel Hill) in 2002, went to grad school (U Michigan) in 2006, and joined Microsoft in 2012. It's been a fascinating, challenging, and rewarding time. You have contributed to my success, my knowledge, and my joys. Thank you for all the great memories! 😊 Here are seven points of unsolicited advice from an old computer / data scientist. YMMV. Read at your own risk. 1. At work, always prioritize IMPACT. Nobody cares how hard you work unless it drives major impact. Choose projects that you love and that are likely to succeed. If they fail to work, or fail to make impact, YOU will be responsible. It may not be fair, but it's what happens. 2. Prioritize family and friends. Our jobs give us the ability to support our loved ones and I'm grateful for how I've been able to help mine. OTOH, I'll never get back those days and nights I spent working over vacations. It helped my career, but it wasn't worth it. Nobody ever said on their death bed: "I wish I had worked more." 3. Prioritize your health. Life is short. You will all be my age sooner than you think. I'm retiring so I can enjoy my life while I still have good health. 4. There is MUCH more to you than your work. Many of you worked hard in school, then university, then your job, always pursuing the next level. You sacrificed so much to earn your job and position. But have you truly taken the time to learn about YOU and the larger world? The people who cook our food, clean our offices, and drive our Ubers may not have our financial wealth. But many have life wealth far beyond what we could ever imagine. Assume they know things that you don't. Are you experienced? 5. To my foreign-born colleagues: You make us better. We are lucky you are here. I know there's a lot of anxiety due to the current administration. We have seen the emergence of a lot of bigotry. But please remember there are many -- I believe the majority -- who support you and recognize that you are HUGE contributors to our society and economy. I've learned so much from you sharing your culture and perspectives. BTW, ahem, I'm Canadian, eh? 6. Learn more about where you live. If you are in Washington, you are on Native land. Proud people lived here before us, and many still do. It's easy to be blind to the history, but it's important to learn and this will make you a richer person with a deeper understanding of the land where you walk. Try to learn at least one word in the language of this land. In Seattle, it's the Lushootseed language. You can learn more at lushootseedresearch.org or at the Hibulb Cultural Center on the Tulalip Reservation. 7. Watch Office Space (movie). Don't let them turn you into a Milton. 😁 Thank you, and all the best!!!

What do you do after you bomb a Zoom call? A) Close the curtains, turn out the lights, eat a pint of ice cream. B) Replay it at 2am, director's commentary included. C) Fake your own death. Move to a small town. Take up goat farming. Or... you could get some coaching. Huckleberry (everyone's executive coach) can debrief any call, the moment you hang up. What landed, what didn't, what to do differently next time. Privately. In about two minutes. Talk it through. Save the ice cream for a real emergency. PS: Thank you Ritika, Anna and the team at the NYU Coaching and Technology Summit for having us in New York. 🍎

Most code was crap before Vibe Coding, just unseen. Ai just makes this more democratic. Anyone can create bad code now.

By the time you finish reading this sentence, a frontier model has scanned an open source project and handed back a list of critical vulnerabilities. And the patch rate on findings like those are still in the low single digits. Today we announced that Cisco, along with Amazon Web Services (AWS) Anthropic Google IBM JPMorganChase Microsoft OpenAI and many more key players, is a foundational member of Akrites launched by the The Linux Foundation, to solve for this impedance mismatch. A serious vulnerability in a major open source project used to take a human expert weeks to find. A model now does several in minutes, several in a single pass. Discovery and exploit have collapsed into the same instant while patch and deploy still run at human speed, on maintainers who don’t have the support to outrun a machine. And this is exposure for everyone - every app deployed, every piece of infrastructure, every vendor's products - the products we all use everyday - are built on the same open source (refer to the well known xkcd comic). So no vendor's walls are high enough to make this someone else's problem. A misaligned set of efforts scanning the same open source just buries the maintainers under duplicate reports, and raises the odds things leak before there is a fix. The only response that matches AI-native detection speed is a an AI-native but coordinated one - one path upstream, maintainers in control, with the fix flowing back into each project's own home, and that is exactly what Akrites is trying to deliver. We, and everyone else, are participating in Akrites as builders - bringing our engineering and open source work to help solve for this need. We would love everyone reading this to jump in and help in the effort as well. Read the official charter and find out how to get involved here: https://akrites.org/