Thomas Markovich Machine Learning Research Scientist Machine learning research scientist. I write about the systems we build and what it means to build them.

The Org Chart Is Not the Work

Companies automate their org charts and call it transformation. The failure mode is organizational. The company automated an artifact of its own political history, a capable enough model pointed at the wrong target, and mistook that artifact for work.

Read more →

On Not Extracting

The obvious question is “Sure, but you build these systems – what does it mean to practice Gelassenheit from inside the apparatus you’re describing?” I don’t have a clean answer; however, in sitting with it, I at least have a clearer sense of why it’s hard to answer. The practice erodes the very capacity for genuine...

Read more →

Everything as Resource

I had an intern ask me today how we maintain our humanity in the age of AI. It’s a simple enough question, and one that comes from an honest place when you spend your days building these systems. I gave what felt like a reasonable answer: I outlined how important it was to engage with friends and family, with litera...

Read more →

Rethinking My Relationship With Technology

I’ve been recently struggling with many little questions that have ultimately culminated in one, big, question. It’s been deeply frustrating, because the only answer that I keep coming to is “I don’t know.”

Read more →

Modeling Life Outcomes with Stochastic Methods

IntroductionMy wife and I got married a little less than a year ago, and we’re starting toconsider the life that we can build together. Like most couples, we’ve beentrying to think through activities such as starting a family, buying a home,and possible career changes; and how to make sure that we can achieve theset...

Read more →

Croissants and Macarons

Not all problems in life can be solved by machine learning. Sometimes you wantto bake something extremely unhealthy and extremely delicious; and I’m fortunateto have the free time to pursue these desires. Recently I learned how to make,but not pronounce, croissants and macarons through lots of trial and errorand som...

Read more →

NeurIPS 2018 Recap by Forge.AI

With quick reflexes and a fortunate server error, I was lucky enough to get a ticket to the 2018 Neural Information Processing Systems Conference (NeurIPS). It was with great excitement that I attended to represent Forge.AI this year. NeurIPS provides its attendees with a week of talks, demonstrations, and incredibl...

Read more →

Knowledge Graphs for Enhanced Machine Reasoning at Forge.AI

IntroductionNatural Language Understanding at an industrial scale requires an efficient,high quality knowledge graph for tasks such asentity resolution and reasoning.Without the ability to reason about information semantically, natural languageunderstanding systems are only capable of shallow understanding. As there...

Read more →

Linking Python to C with CFFI

Why link Python to C?I hope that it is uncontroversial to state that Python is a great language that can suffer from occasional performance issues. This is especially true if Python is being used in heavy numerical computing environments like those at Gamalon. Gamalon is not the first to require using Python for num...

Read more →