295: A Deep Conversation About Tech & Life

Episode Summary

Executive Summary: Kirill and Adlan de Pontavice discuss Adlan’s seventh appearance on the podcast, his personal updates, and his upcoming Data Science Go 2019 workshop on deep NLP and BERT. They also cover reinforcement learning, RPA, career-soft skills, AI’s impact on jobs, and personal development themes like cold showers, meditation, and life dreams.

Main Topics: Adlan’s personal update and life rhythm: Adlan shares that he attended three weddings in one week, traveled to Vienna, and is balancing work, travel, and recovery. The conversation uses this to reflect on how intense life events make time feel faster. Data Science Go 2019 and Adlan’s BERT workshop: A major focus is Adlan’s upcoming DSGO session on deep NLP, specifically Google’s BERT model. He explains why he chose NLP, how BERT differs from seq2seq approaches, and that the session will be hands-on in Google Colab. Reinforcement learning and TD3: Adlan discusses his recent deep reinforcement learning course and the TD3 (twin-delayed DDPG) algorithm, describing it as a state-of-the-art method that combines Q-learning and policy gradients and trains efficiently in complex environments. RPA as a growth area for business and training: Both speakers discuss robotics process automation as an important, rapidly growing trend. They see it as easy to implement, useful for eliminating repetitive work, and a good opportunity for new courses, certifications, and client projects. AI, jobs, and the need for retraining: They debate automation’s social effects, arguing that while jobs will disappear, many more new jobs will be created. Their view is that the main challenge is preparing workers through training rather than resisting automation. Soft skills, presentation, and communication: Kirill’s DSGO workshop focuses on presentation and communication skills, which he argues are increasingly valuable because technical tools are becoming easier to use while human communication remains a differentiator. Personal goals, love, and inner peace: Toward the end, the conversation turns reflective: Kirill wants greater inner peace and less judgment; Adlan says his dream is to fall in love again. This closes the episode on personal growth and connection.

Key Arguments: BERT is a major leap in NLP because it outperforms previous state-of-the-art approaches like seq2seq and does so largely through attention-based architecture rather than encoder-decoder structure. Deep NLP is worth teaching now because it is one of the most disruptive AI subfields and BERT represents a rare quantum leap rather than a gradual improvement. TD3 is significant in reinforcement learning because it is fast to train, combines multiple RL techniques, and achieves state-of-the-art results on benchmarked tasks. RPA is an especially practical business opportunity because it automates repetitive tasks using accessible tools like UiPath and is easy to adopt relative to many AI projects. Automation fears are often overstated; the better framing is that jobs will transform, many new roles will emerge, and training should prepare people for the transition. Soft skills such as presentation and communication are becoming more valuable as technical implementation becomes increasingly accessible through tools like TensorFlow 2.0. Hands-on learning is central to both workshops and courses; Adlan’s BERT talk will use Google Colab and Kirill’s workshop will involve practice and feedback. The speakers view AI as useful for high-impact domains such as healthcare, including preventing heart failures and strokes through predictive applications.

Data Points: Podcast episode: 295 - This is the episode number introduced at the start of the transcript. Adlan’s appearances on the podcast: 7th time - Kirill notes this is Adlan’s seventh appearance on the Super Data Science Podcast. Weddings attended by Adlan in one week: 3 - Adlan says he attended three weddings in France in the same week. Trip to Vienna: 1 round trip - Adlan says he made a round trip to Vienna in the middle of the wedding week. Cold showers practice: 2 weeks - Adlan says he has been doing cold showers for two weeks. Blue Life leads per week: 3 new leads weekly - Adlan says the business generates about three new interested leads each week. Emails sent via LinkedIn: 20 to 30 people - Blue Life outreach is described as sending LinkedIn emails to 20–30 people. DSGO attendance 2017: 150 people - Kirill cites 2017 attendance for Data Science Go. DSGO attendance 2018: 350 people - Kirill cites 2018 attendance for Data Science Go. DSGO expected attendance 2019: 600 to 800 people - Projected attendance for the 2019 conference. Countries represented at DSGO: 23 countries - Kirill says attendees come from 23 countries worldwide. DSGO dates: 27th, 28th, 29th of September - The conference dates announced during the episode. Data Science Go episode count: 3rd year - Adlan is returning for the third consecutive year. Students/community size: 900,000 - Kirill says the student base has just crossed 900,000. Future student goal: 1 million by end of 2019 - Kirill expresses hope to reach one million students by year-end. RPA growth forecast: 800% in the next five years - Adlan references strong projected growth for RPA. Job displacement estimate: 70 million jobs - Adlan cites a figure for jobs that may disappear due to automation. Job creation estimate: 130 million jobs - Adlan cites a larger number of jobs expected to be created. South Dakota event timing: 10 days - They say they will meet at Hero Club in about ten days. Road trip drive time: 28 hours over 2 days - They discuss a planned drive to DSGO and estimate the total trip as 28 hours split over two days.

Pivotal Quotes: "BERT, B-E-R-T, the BERT model that was released by Google... took everything in terms of natural language processing to a whole new level, like a whole quantum leap in that space." — Kirill Aramenko: Kirill explains why Adlan’s upcoming DSGO topic is exciting and important. "This year, 2019, will be very different because this time I won't be giving a speech on a general topic, I will actually give a training on deep natural language processing." — Adlan de Pontavice: Adlan describes the format and focus of his 2019 Data Science Go session. "I think 40 million jobs will disappear by 2023... 70 million will disappear and 130 million will be created." — Adlan de Pontavice: Adlan discusses automation’s impact on employment and the need for retraining.

Implications: Listeners should expect NLP and automation skills to matter more, but communication and adaptability may matter just as much. The episode frames AI as both a technical and human transition, with strong demand for hands-on learning and retraining.