#60 Building Self-Driving Cars in Silicon Valley with Vladimir Iglovikov, Ph.D. – Senior Computer Vision Engineer and Kaggle Grandmaster

vlad.jpeg

Vladimir Iglovikov graduated from university with a degree in theoretical physics, he moved to Silicon Valley in search of a data science role in the industry. This led him to his current position in Lyft’s autonomous vehicle division where he works on computer vision related applications. In the past few years, he has invested a lot of time in Machine Learning competitions leading to his title of Kaggle Grandmaster.

In this episode, Vladimir explains how difficult it was to find work in Silicon Valley. He had harsh requirements for a salary, no one looked at his resume. Companies in Silicon Valley are willing to pay big bucks, but at the same time, they require the person to be skilled in software engineering, machine learning, and statistics. His biggest issue when applying for jobs was assuming that all people are similar to the people in academics. At his interviews, he felt no connection with the interviewers. After sending his resume to over 200 different companies, someone finally bit just before his visa expired. Vladimir worked at Bidgely for 8 months then moved to TrueAccord and eventually got his job at Lyft. 

As a manager, you need to learn to communicate and excite people in different teams about your project. When people come from academia, they are smart on paper but may only get a mid-level job. People with soft skills, like management and marketing, will be more impactful for the company. Around four years ago, Vladimir started with Kaggle. He needed to do something to help apply his skills. He compares working with Kaggle like working out at the gym. You lift weights, but you may not necessarily lift weights when you are outside of the gym. However, it does impact your appearance, strength, posture, and confidence. So, Vladimir started participating in Kaggle competitions and epically failing. Every competition helped Vladimir strengthen a particular skill. Machine learning is an applied discipline, Kaggle teaches this really well. 

Later, Vladimir touches on hiring data scientists. He says nobody knows how to hire. In Silicon Valley, so many people are competing. If you are performing well in the company, you get promoted. However, you can invest in looking around and finding another job where you can get a raise. These companies are also looking for an applicant with at least 10 years of experience when, in reality, they just want a nice person who can code. Stay tuned to hear Vladimir discuss machine learning, autonomous driving, and why he is not in a rush to pursue his own startup. 

Enjoy the show!

We speak about:

  • [02:00] How Vladimir started in the data space

  • [12:30] Transferring from academia to industry 

  • [21:40] Benefits of having soft skills   

  • [25:45] How Vladimir manages stress 

  • [31:30] Kaggle is like lifting weights 

  • [35:30] The hiring process for data scientists  

  • [40:45] Excitement for machine learning 

  • [46:00] Autonomous driving   

  • [47:55] Pursuing a startup

  • [51:40] Aiming to maximize mistakes in a day 

  • [61:00] Social life comes first 

Resources:

Vladimir’s LinkedIn: https://www.linkedin.com/in/iglovikov/

Kaggle: https://www.kaggle.com/iglovikov

Quotes:

  • “When you are coming to work at a startup, you have to wear a lot of hats.”

  • “On average, people work in Silicon Valley 1.5 years.”

  • “I encourage everyone to move to Silicon Valley for some time while you are young.”

  • “Building a startup is a rare and complex job.”

Books:

The 48 Laws of Power

So Good They Can't Ignore You: Why Skills Trump Passion in the Quest for Work You Love

The Making of a Manager: What to Do When Everyone Looks to You

Mastery

Now you can support Data Futurology on Patreon!  

https://www.patreon.com/datafuturology 

Thank you to our sponsors: 

UNSW Master of Data Science Online: studyonline.unsw.edu.au 

Datasource Services: datasourceservices.com.au or email Will Howard on will@datasourceservices.com.au 

Fyrebox - Make Your Own Quiz!

Vladimir Iglovikov is based in San Francisco, California.

And as always, we appreciate your Reviews, Follows, Likes, Shares and Ratings. Thank you so much for listening. Enjoy the show!



#32 Carole Wai Hai - Head of Data Science & Analytics

Ep32 Carole Wai Hai - Berlin, Germany.jpg

Carole had an unusual path into data science. She's worked as a content project manager, in strategic planning and in sales before getting into data through Business Intelligence at Fyber where she eventually became their Head of Analytics. Today she is the Head of Data Science & Analytics at Tenjin.

We speak about:

* The strengths of being a generalist

* Upskilling throughout your career

* Focus on self service reporting

* The skills needed in a BI team

* Creating internal user groups to share knowledge

* Convincing people to get training on the tools required to do their job better

* The benefits of gaining a reputation internally

* Setting a strategy for data teams

* The importance of data modelling skills in data teams

* Learning technology on the job when you're background is not technology

* Monthly meeting with key departments to review all dashboards in the department

* Working remotely in global companies

* Metrics about user behaviour

* Offering analytics for many customers with the same problem/need

* How to develop consulting skills

* The platinum rule - book on communication style

* The leadership challenge - book recommendation

* What it's like working in startups

* How to recover from being a workaholic

Carole is based in Berlin Area, Germany

And as always, we appreciate your Reviews, Follows, Likes, Shares and Ratings. It really helps new data scientists find us. Thank you so much, and enjoy the show!

Thank you to our sponsors: 

UNSW Master of Data Science Online: studyonline.unsw.edu.au 

Datasource Services: datasourceservices.com.au or email Will Howard on will@datasourceservices.com.au 

Fyrebox - Make Your Own Quiz!

#29 Dr. Klaus Ifflander - Chief Analytics Officer

Ep29 Klaus Ifflander - Berlin, Germany.jpg

Klaus started his career doing internships at Yahoo! and the port of Hamburg. He worked as a consultant and completed a PhD in Quantitative Marketing. Today he is the Chief Analytics Officer at YAS.life

We speak about:

* The importance of getting applied experience as early as possible

* Defining KPIs for businesses

* Using data to change organisational behaviour and increase safety

* How to navigate organisations to create data definitions

* Realities of consulting: positives and negatives

* Why large companies require so much custom work

* How to help people and organisations that don't know what they want

* Helping organisations in progressing through their analytics journey

* How to overcome technical challenges with creative solutions in your projects

* Why honesty within yourself and others is imperative in your work

* How to provide customers what they need instead of what they want

* The importance of hard and soft metrics when measuring value

* Applying soft skills in data science

* How to find what will be valuable for your customers

* Expanding your interest with a postgraduate degree

* How your social surroundings affect your purchase decisions

* Using soft skills for data acquisition

* What is eigenvector centrality and what is it used for?

* How product reviews influence your buying decisions

* How to create experiments in business

* Pricing models in the steel business

* Data science in fitness startups

Klaus is based in the Berlin Area, Germany.

And as always, we appreciate your Reviews, Follows, Likes, Shares and Ratings. It really helps new data scientists find us. Thank you so much, and enjoy the show!

#28 Jennifer Prendki - VP of Machine Learning & Data Strategist

Ep28 Jennifer Prendki - Mountain View, CA.jpg

Jennifer started her career as a particle physicist before becoming a data scientist. After gaining experience in many fields including high frequency algorithmic trading & advertising, she was Atlassian's first Chief Data Scientist. Today she is the VP of Machine Learning at Figure Eight and an Expert and Advisor at the International Institute for Analytics.

We speak about:

* How to see the results of your work sooner and faster

* The importance of choosing your manager

* Making data strategy decisions for companies that are very immature in their approach to data

* Building data science teams from scratch

* Combining impostor syndrome and leaps of faith for your benefit

* The importance of making mistakes to be successful

* What having a great data culture really means

* How to convince peers and supervisors on the benefits and the path of data strategy

* Differences between having a technical and non-technical manager

* Combining technical abilities and business sense

* The importance of customer contact for technical people

* Focus on the impact and outcome of everything that you're building

* How to keep the balance in teams

* Pleasing customers vs product intuition

* How to drive and create a data driven culture

* How to create scale with your data science efforts

* How to build your data science team

* Data engineering vs Machine learning engineer

* How to keep talent

* How can data scientists learn the skills for business leadership

* Active learning and building products for data scientists

Jennifer is based in Mountain View, California

And as always, we appreciate your Reviews, Follows, Likes, Shares and Ratings. It really helps new data scientists find us. Thank you so much, and enjoy the show!

#25 Ben Taylor - Chief Al Officer & Cofounder

Ep 25 Ben Taylor - Provo Utah Area.jpg

Ben started his career as a chemical engineer. He developed an interest for computer vision early on. He worked for Intel, then at a hedge fund and then became the Chief Data Scientist at HireVue. A couple of years ago he started his own AI startup called Ziff.ai where he's is building a Deep Learning platform for product visionaries and software engineers.

We speak about:

* How computers amplify us

* What it looks like to start your own AI company

* How to switch programming languages

* Downsides of Google's tensorflow

* What industry expects from data science

* How to deliver value with ML

* How to pick ML projects to tackle

* Eliminating bias in AI applications

* AI powered job interviews of the (near) future

* Topic discovery with DL

* AI warfare in business

* What is a Hive Mind and how it works

* Future health care assessments at home

* AI is cute until it's scary

* The importance of passion and obsession in data science

Articles by Ben on Linkedin:

This is Why Your Data Scientist Sucks:

https://www.linkedin.com/pulse/why-your-data-scientist-sucks-benjamin

The Al War Machine: Our Darkest Day

https://www.linkedin.com/pulse/ai-war-machine-our-darkest-day-ben-taylor-deeplearning-/

The Al War Machine: The Hive Mind

https://www.linkedin.com/pulse/ai-war-machine-hive-mind-ben-taylor-deeplearning-

Getting That Data Science Job

https://www.linkedin.com/pulse/getting-data-science-job-ben-taylor-deeplearning-/

From 0 to $100K+ data science job in 6 months

https://www.linkedin.com/pulse/from-0-100k-data-science-job-6-months-ben-taylor-ai-hacker/

Ben is based in the Provo, Utah Area

And as always, we appreciate your Reviews, Follows, Likes, Shares and Ratings. It really helps new data scientists find us. Thank you so much, and enjoy the show!

#23 Mario Vinasco - Marketing Analytics and Data Science Manager

023_mario.jpeg

Mario is an Electrical Engineer from Colombia. He went to Silicon Valley to do his Masters at Stanford University and stayed to build a career in Marketing Analytics. He has incredible experience and has worked at Intuit, Google, HP, Symantec and Facebook. He currently works at Uber as Marketing Analytics and Data Science Manager.

We speak about:

  • Starting in marketing analytics without knowing anything about it

  • The creatives and the quants of marketing analytics

  • How the headaches of tech have changes over the years for data scientists

  • Data dictators and why multiple versions of the truth are necessary

  • How to communicate results of your analysis to executives

  • The importance of data science education in organisations

  • How to pick the best predictive model for your applications

  • Analytics as the art of counting

  • The importance of working with inspiring people

  • Why running experiments is the gold standard of finding what works

  • How to use people analytics - Google style

  • Why your job is to empower your stakeholders - and how to make this a 2 way relationship

  • How to not saturate your channels

  • What he looks for in CVs and applicants

  • How to stand out during interview processes

  • The coolness of network partitions at Facebook

  • Using embeddedings at Facebook - using vectors when one hot encoding gets too cumbersome

  • The importance of rapid prototyping

  • The value of collaboration across your organisation and what that looks like in practice

  • Things to look for and ask during your interview process

  • How to choose where to work

  • Setting up new practice globally at Uber

  • Product manager as a facilitator, protector and enabler



Mario is based in the San Francisco Bay Area

And as always, we appreciate your Reviews, Follows, Likes, Shares and Ratings. It really helps new data scientists find us. Thank you so much, and enjoy the show!

#20 David Greenberg - Senior VP and Head of Data Analytics

020_david.jpeg

David studied applied physics and began his career as a consultant. He’s had his own company where he created a video asset management & workflow software in the 90s!. Then worked in the education/not-for-profit sector and then went into the finance sector as VP of BI & Data Analysis. Today, he is the Senior Vice President and Head of Data, Analytics and Research at BankMobile.

We discuss:

  • the insights into large companies from his early days in consulting

  • why technology provides the “guard rails” for the business

  • why our roles as data scientists is to make sense of the mess

  • what’s missing in today’s analytics education and how to learn what you need

  • what to look for when building a diverse team

  • the importance of creating a narrative in analytics

  • the mindset to maintain during your analysis

  • motivations behind problems with data definitions

  • how data is like a flashlight

  • how analytics professionals help companies to judge performance fairly

  • how to create a data-driven culture in your organisation

  • what to do when companies don’t know their basic metrics

  • what questions to ask when you receive data requests

  • issues in horizontally integrating data among very vertically structured organisations

  • lessons learnt from his tech startup

  • the importance of being honest and kind

  • how to reframe what could be seen as attacks in order to be kinder and more understanding towards others

  • how to hire the best people out there when your organisation can’t pay as well as others

  • tips on navigating corporate politics

  • who should own and be involved in building the data warehouse for the organisation

  • benefits of having a centralised analytics team

  • why, as analytics professionals, we’re entering the world of expectation management

David is based in New London/Norwich, Connecticut Area

And as always, we appreciate your Reviews, Follows, Likes, Shares and Ratings. It really helps new data scientists find us. Thank you so much, and enjoy the show!

#18 Dr. Ahmed Khamassi - Vice President of Data Science

018_ahmed.jpeg

Ahmed started his career at Siemens research, worked in startups, at Google, PayPal, SAS and JPMorgan and has his own machine learning company and now he runs data science at Equinor

We speak about:

  • benefits of simulations in research and data science

  • how he went from “equation-driven” to “data-driven”

  • the role of simulations in optimisation, decision-making and automation

  • uses of simulations and deep reinforcement learning models in the energy industry

  • how data is used 2-5kms underground below the sea to infer the properties of the ground underneath

  • lessons from startups and what to look for in people to work with

  • why it’s important for data science teams to own the engagement of value creation with the customer

  • how to ensure that your data science team is creating value in your organisation

  • how to prioritise the work done by your data science team and what to aim for

  • how to make your role and your teams’ roles redundant by delivering disproportionate amount of value

  • what to aim for with data driven products

  • what startup mentality means and how to bring startup DNA into a traditional business

  • how to engage business stakeholders to get SMEs working in your team

  • the role of the business product owner and what to look for when picking one

  • how to get the best out of your team

  • lessons learnt at Google and his aspirations for his current team

  • why the new style of innovation management cannot be used pervasively in all businesses - how businesses wins

  • operational excellence vs innovation in different industries

  • how to create a fantastic organisational culture

  • problems BI tools should be looking to solve

  • The importance of dev ops in the data science value chain

Ahmed is based in London, UK

And as always, we appreciate your Reviews, Follows, Likes, Shares and Ratings. It really helps new data scientists find us. Thank you so much, and enjoy the show! 

-------------------------------

Also, catch Felipe at the Chief Data & Analytics Officer Conference in Melbourne on September 3-5, 2018

https://chiefdataanalyticsofficermelbourne.com/

#14 Dr Gabriel Maeztu - Medical Doctor, Co-Founder & Chief Data Scientist

014_gabriel.jpeg

In this episode we speak with Gabriel Maeztu who is the Co-Founder and & Chief Data Scientist at IOMED Medical Solutions. We talk about:

- his background, how he went from medicine to data science and how he combines medicine, data science & entrepreneurship
- how to start coding when everyone around you tell you you’re crazy
- image processing in medicine, using scikit learn to classify patients
- how to use data science to validate what you’re taught in medical school
- economical Incentives of the medical system that is probably slowing down progress in the data space
- GAFAs: Google Apple Facebook Amazon in medical data
- value based care built on data science
- NLP/text processing in medicine
- current & future data challenges in medicine and much, much more!

Gabriel tells us about his company IOMED Medical Solutions (https://iomed.es/) where they have built an intelligent scheduling system for outpatient clinics that uses the patient history and their likelihood to have or get certain diseases in order to prioritise their visits and assist doctors in providing better care.

When discussing Electronic Health Records (EHR) also known as Personal Health Records PHR we talk about a couple of standards that are making it easier:

We talk about international file standards that are trying to address interoperability issues in medicine such as:

  • Health Level-7 or HL7 which is a set of international standards for transfer of clinical and administrative data between software applications. We talk about this regarding the interoperability in medicine (http://www.hl7.org/)

  • Fast Healthcare Interoperability Resources (FHIR, pronounced "fire") is a draft standard describing data formats and elements (known as "resources") and an application programming interface (API) for exchanging electronic health records. The standard was created by the Health Level Seven International (HL7) health-care standards organization. (http://hl7.org/fhir/)

As you’ll hear in this episode, there’s A LOT of opportunities to work on this field for anyone interested. Gabriel is very open and honest with the challenges in this field and the ones that his company has faced and overcome. I’m very thankful that he was able to share i this way.


IOMED is hiring data scientists! 
angel.co/iomed/jobs/379740-data-scientist 

Show notes: www.datafuturology.com/podcast/14 

Gabriel is based in Barcelona, Spain

#12 Alessandro Pregnolato - Director of Analytics

12_alessandro.jpeg

In this episode we speak with Alessandro Pregnolato, he is the Director of Analytics at Typeform.com. We talk about:

- his journey to get where he is, 
- what is the optimal size of a data science team, 
- how to use data science in SaaS businesses/startups
- the 4 pillars of a great data strategy
- how to be an expert generalist in the data space and much more!

Alessandro is a Business Analytics Leader with a love for Data Science. He has over fifteen years experience within the domain of BI, Analytics, Big Data and Machine Learning in international environments. 
He has strong management and communication skills with a demonstrated ability to work well under pressure with people from a variety of backgrounds.

Show notes: www.datafuturology.com/podcast/12 

Alessandro is based in: Barcelona, Spain