Scott Draeger on delivering AI

Brian and Don welcome a much anticipated guest for this episode, Professor Fernando Perez joins us for an episode of 26.1 AI Podcast. Dr. Perez speaks about his journey, the community, and all the challenges along the way. Fernando shares in his inimitable style, how he journeyed from straight laced physicist in pursuit of an academic career to doggedly ignoring naysayers and creating one of the most important components of the modern PyData stack. One personal challenge during this journey was losing a friend Dr. John Hunter. John also influenced your host Brian Ray. Though John missed collaborating when Fernando set out with IPython because of conflicts from prior commitments, the two joined in later to collaborate on advancing tools data scientists use every day now. Sit back and enjoy Fernando's dexterity on multiple topics as hosts Brian Ray and Don Sheu hold on for the ride for your benefit listener.

[Part 1 of 2] Listeners join in for a wonderful conversation in this episode. Our guests Matthew Rocklin and Hugo Bowne-Anderson are extending access to powerful distributed computing for more data users with their startup Coiled (https://coiled.io/). Data scientists with a two minute download of Coiled’s software (https://cloud.coiled.io/) can scale their work to the cloud. We discuss during the episode how conversations with the open source community resembles early customer conversations commonly used by entrepreneurs in a lean startup framework. Dask’s creator and Coiled founder Matthew described his software design approach that has a decided minimalist bent. A great benefit for users of popular Python libraries because of Matt’s approach is a familiar interface when using Dask or Coiled to extend the power of popular PyData stack tools. Our conversation turns to how Coiled has the capability to extend more computation power to many casual users of Python who are interested in solving data problems pragmatically without rebuilding a data factory every time. [Join us next week for Part 2]

Tileli Amimeur changed direction from a fast track traditional software career based on intriguing invitation to apply coding to biology. This shift followed study in undergrad and grad schools at University of Texas, and a string of jobs with top notch traditional technology companies. Now Tileli is happily applying computation in search of important cures with an interdisciplinary team of AI experts, biologists, and chemical engineers. A must listen no matter your day job. Listen for an understanding of how AI is touching every point of human experience, including coding proteins for cures.

Our prior guest e2f CEO Michel Lopez introduced this week's guest. Intento founder Konstantin Savenkov walks us through a tutorial on how to buy and sell AI. A dense episode worth a listen both for anyone considering buying AI for an enterprise, and founders seeking to sell an AI product to buyers. Intento serves companies making sense of 32 different possible translation vendors. Intento identified the complexity presented to buyers by such a large population of possible vendors as a good market opportunity for building a company. For startups seeking to find product market fit, Konstantin shares step by step the series of decisions that led Intento to concentrate on helping companies with selecting machine translations vendors. We discuss some practical advice on pitfalls many enterprises encounter trying to force procurement process built for buying traditional software onto decisions on new work processes created by AI products.

Our guest Kingsley Ndoh, MD has a global health project centered in Ondo state in Nigeria, where he endeavors to collect the largest set of cancer related data on people of African descent. We discuss during the episode, before algorithmic bias, how biased decisions about data collection has fatal consequences for cancer patients of African descent. Dr. Ndoh speaks about how Black Panther star Chadwick Boseman was far younger than current guidelines in the United States for early screening of colon cancer. Perhaps guidelines would change for the age to start early screening of cancer if we had the data on diverse populations. Listen all the way through for a consequential discussion that may help guide you as you practice your profession. Dr. Ndoh started out life in Jos, Nigeria. After training as a physician, he pursued his systems level interest in healthcare with further study at the University of Washington. He completed a masters in public health at UW, and since completing the degree he's joined the faculty.

Our guest for this episode shares with us the challenges their firm e2f faces when collecting data for training voice AI applications. Michel Lopez started early on and researched AI and taught the subject in Saudi Arabia and Thailand before the advent of cloud. Instead of practicing AI following teaching, he took a left turn into applying his multilingual skills to performing translations for technology companies. Serendipity brought Michel a new opportunity when a translation customer asked for help collecting voice data. Voice is our computational interaction for the future. Take a peek on how we're getting to this future in this latest episode of 26.1 AI Podcast

Our guest this week Shreya Nallapati founded an organization to help prevent mass shootings. From the #NeverAgainTech website, a quote from Shreya explaining her passion for this project, "I founded this organization after hearing Emma Gonzalez's powerful speech. Growing up with Columbine, I was tired of hearing friends and family being impacted, without being able to do anything about it. So I naturally took the areas I am specialized in, artificial intelligence and data mining. Currently, over 105 teenagers, policymakers, and industry professionals are participating in this project." Importantly, Shreya contrasts #NeverAgainTech against vendors selling tech to law enforcement. Included in the comparison is what #NeverAgainTech has done to ensure they're not part of the problem with introducing a new form of bias with AI. In this episode, learn more about this rising star in AI. Thank you to our friend Ruthe Farmer of CSforALL and Last Mile Education Fund for introducing this week's guest.

Our guest this week on 26.1 AI Podcast is the long time Open Source Guru Ian Bicking. If you are a Python developer and ever typed, "pip," at the command line you can thank Ian. On this episode, he speaks on voice as a computing interface and some of the etiquette of addressing our assistants. How soon will AI improve to the point of being emotionally manipulative with users? Do you want to grade essays turned in by college students, cheating with GPT-3?

Our host Brian Ray's staff colleague David Law joins with a special treat. David has led development of an open source project titled "data⎰describe". This Python package accelerates exploratory data analysis. David also shares how he transitioned from studying material science at Cornell to completing a masters in data science at Northwestern and practicing as a data scientist at Maven Wave. We manage to add to the episode, how AI experts feel about riding in a self driving car, right now.