When I’m out of office
I’m likely to be found badly singing 1960s-80s rock songs with my kids or doing 3-D printer design and construction.
What would you do if you had more time?
I’d work on my welding and metal fabrication skills with the aim of someday making a pulse-jet powered bicycle.
It was kind of a Blues Brothers moment – Jake (Sri Chandrasekar) and Elwood (Dan Gwak) were looking to get the Lab41 team back together to create Hyperscale. Yonas Tesfaye, Sri and Dan and I had previously worked together there, and I really enjoyed the energy and expertise that each one of them brought. When Sri gave me the call, I knew I would be leaving my then-job to join them.
I have seen and done a lot of interesting things in my career, from applied mathematics, software and hardware security analysis, hardware design and big data analytics to machine learning. All the work was different but was unified around solving unique problems within difficult constraints, and all of it helped prepare me for Hyperscale.
- Facebook I helped combine big data, machine learning, and data streaming to make security decisions; created a pipeline for streaming security events from employee laptops; created a hardware implant to troll our internal red team instead of working on promotion paperwork like I was supposed to be doing; and used X-rays, laser scanning and thermal imaging to verify hardware integrity.
- Lab41 I created a facial recognition pipeline for finding individuals across large volumes of video data. I created Poseidon – software-defined networking feeds machine learning algorithms for training – and network posture updates. I also supported MagicHour, a process that turned semi-structured log data into events that were easier for analysts to read and interpret.
- Government work I solved nation-state problems for multiple intelligence agencies, did rapid prototyping of big data analytics, and worked on machine learning, classification analytics, recommendation systems design and time-series correlation. I also helped teach coworkers about big-data optimization techniques, software design and mapping algorithms to technologies.