In Case You Missed It
Modeling in Place
A panel discussion with Alectio, NVIDIA, Pandora, and Yelp
In today’s challenging sheltered environment, work from home is the new norm for many data scientists and engineers building and tuning models. Today, we hosted a panel for a number of modelers across a wide spectrum of tech companies in Silicon Valley and beyond. In particular, we asked them:
- What modeling infrastructure and tools they like to use
- How to improve productivity in meetings and brainstorms
- How they schedule and time-box their daily activities, across individual work, meetings, and breaks
- Their favorite way to spend their breaks
Here are some quick highlights:
- Jennifer, founder of Alectio, shared how whiteboarding is a challenge over video conferencing, but her best engineering hires are the ones who were able to collaborate effectively without even a conference table in Alectio’s early days. Whiteboarding is a useful tool, but today a luxury, and the top performers can thrive without it! She prefers modeling in PyTorch over Keras, and is grateful that she now has more time to spend with her children.
- Rajan has found an opportunity for his team to re-allocate time spent commuting to building models instead. At NVIDIA, he’s lucky enough to have access to large compute GPU clusters, though oftentimes he’s building demos and proofs-of-concept on customers’ public cloud infrastructure. He’s a big fan of RAPIDS.ai, which will even GPU-accelerate data prep.
- Oriol works on Pandora’s recommendation engine, using deep learning models to assess properties of tracks that correspond with users’ preferences. He’s no stranger to working from home since his team works remotely every Friday, but he advised getting a standing desk for the long-term benefit of his back. He misses “3D people” and the helpful interactions that can only happen in person, and he finds it important to keep work and leisure separate in his calendar. He enjoys getting outside to Oakland’s Redwood Regional Park for a hike and keeping 6 feet of physical distancing while he does!
- Maria, from Yelp, has been working remotely for several years now from the east coast, and enjoys being three hours ahead of her team of engineers. She stresses the importance of working out early in the morning and using the time difference to work on individual work before several hours of meetings. She tries to always find time for coffee or a five-minute break between meetings, and time-boxes her activities to avoid the “rabbit hole” of tweaking a model for hours and hours, when she could have gotten feedback from her colleagues sooner.
- SigOpt’s very own Tobias Andreasen expressed his enjoyment of NumPy over any specific modeling library: sometimes it’s best to express your model first using only the basics. Having spent much of the past year abroad working from Denmark with only a laptop, he admits that he values a more robust workstation at home. Having just moved steps away from the beach, Tobias likes to spend his breaks surfing or taking his recently-adopted puppy for a walk.
Many thanks to Jennifer from Alectio, Rajan from NVIDIA, Oriol from Pandora, Tobias from SigOpt, and Maria from Yelp! We look forward to hosting more modelers and data scientists in the future. You can find the full panel recording on our YouTube channel.
If you’re modeling from home and would like to try SigOpt, please contact us to discuss your needs with our team. If you’re in the academic community and would like free access to our product, please contact us here.