Interview with Data Science expert, Dipanjan Sarkar
What are the advantages of Constructor Learning’s Data Science Bootcamp?
“Having a passion for helping people kick start their own journey in Data Science, I have worked with diverse companies and universities to help build course curriculums to enable people to be job-ready by gaining relevant skills in the area of Data Science. There are several advantages of doing Constructor Learning's Data Science Immersive Bootcamp. First and foremost, which resonates with my personal motto, Constructor Learning believes in a live cohort-based learning system, which means you have a small focus group of 10 to 20 students who interact in live sessions with instructors who teach them relevant concepts covering a breadth of areas in Data Science including understanding statistics, Python fundamentals, linear algebra, databases, Machine Learning, Deep Learning, Natural Language Processing, Computer Vision and even elements of Machine Learning Engineering. This includes deploying models and building end-to-end applications. The instructors are all experts from the industry, and they themselves have been Data Scientists or Data Science Leads for several years. Constructor Learning also brings in guest lecturers from diverse domains in the industry which gives the students exposure to how Data Science is done in the real world.
Another interesting aspect which I find valuable is that it's not just about learning concepts, but also applying them by working on real-world data sets and interesting applications in Data Science. They also work on exercises to reinforce their learning and in case they get stuck, they get direct support from teaching assistants as well as from the instructors themselves. This is an especially important piece of learning because in traditional online course formats you have assignments and live hands-on coding notebooks, but usually, it is never live. It's always pre-recorded and there's no communication between the instructors and the students. Constructor Learning's way of learning is more interactive. Another interesting aspect is that once you reach the end of the program, you get to apply all your learnings in an actual industry Capstone project, done in collaboration with companies from Constructor Learning's network. This is very important because a lot of online courses just end with a certificate and that's it. You don't get to work with a company on a real-world problem and I think this kick starts your journey into Data Science in the right direction. Having a project in your portfolio that you have done based on a real-world industry project is a big advantage. Other interesting aspects include having access to the Swiss job market, which I think is very challenging to navigate on your own. Constructor Learning's wide network of industry-based experts, companies as well as having a wide variety of activities like hiring days and other interesting events, meetups, etc, guides you towards your new career.
The curriculum of the Data Science program is state of the art and constantly evolving over time, because as you know the whole field of Artificial Intelligence, Machine Learning and Data Science itself is ever evolving with technologies being replaced and new techniques coming in. Students learn the latest and the best in terms of concepts, methodologies, tools, frameworks and so on.”
Why should you learn Python?
“Python is a programming language which draws-in elements from Functional Programming as well as Object-oriented programming languages, making it a very powerful general purpose programming language which can be used to solve a wide variety of problems in Data Science, Scientific Computing, Data Analysis, Data Engineering as well as build end-to-end applications and leveraging principles of software engineering. If you're serious about a career in Data Science or Machine Learning Engineering, you should definitely learn Python, given its active developer ecosystem, constantly releasing updates and enhancements as well as an open-source ecosystem of libraries and frameworks dedicated towards leveraging Data Science capabilities in Python. This induces libraries like NumPy, Pandas, scikit-learn, TensorFlow, PyTorch, …
This is why at Constructor Learning, we focus on not just understanding the fundamentals of the Python programming language, but also leverage hands-on examples to understand and implement how to write efficient phytonic code and also master all these dedicated libraries around the Data Science ecosystem. This is fundamental for building your own applications in Machine Learning and Data Science.”
How should you start with Deep Learning?
“Deep Learning is a subfield of Machine Learning, which leverages neural networks to solve increasingly complex problems in areas like Computer Vision and Natural Language Processing. Popular applications of Deep Learning include image classification, object detection, object and instance segmentation, image generation, video classification or detection as well as problems in the area of NLP just like text classification, text generation, chat bots, information retrieval, search engines and many more. If you are interested in getting started with Deep Learning, I believe you should have a solid grasp of Machine Learning fundamentals as well as an understanding of how to build your own models, how to tune and evaluate them based on a wide variety of evaluation metrics. Start with the fundamentals of understanding how neural networks are built including what is a neuron, what is a hidden layer, what are non-linear activation functions, what is a loss function, how gradient descent and backpropagation work. Coupled with the theoretical concepts you should also try to pick up a language like Python along with a library like TensorFlow or PyTorch. Try to build your own Deep Learning model starting with a simple dense or artificial neural network. Play around with the model, the layers, the activation functions and then dive into more complex models like convolution neural networks, require neural networks, LSTMS, GRUs, sequence to sequence models and even pick up more complex models like generative models and transformers. In the next step you should also try to apply these models in different applications like segmentation, classification detection, generation anomaly detection and so on.
At Constructor Learning we firmly believe that students should go through a solid understanding of Machine Learning principles. With this knowledge, they can dive into the essentials of Deep Learning including the various components of a neural network.”
What is NLP and why should you use it?
“Natural Language Processing, popularly known as NLP, is an interdisciplinary field drawing in elements from Computer Science, Artificial Intelligence as well as applied linguistics. NLP focuses on trying to make machines learn, understand and comprehend unstructured text data or audio speech data to perform complex tasks. If you want to differentiate yourself and stand out from other applicants, Natural Language Processing is definitely a skill you should try to master, given that over 80% of data is unstructured in most enterprises today. They are definitely trying to figure out how to tap into that vast reserve of data to extract additional insights, besides leveraging databases, and relational and structured data by looking at techniques like SQL and so on. To master NLP you should try and first understand how to load text data, how to process and clean it, and then look at elements of representation to consume and understand numerical formats which can be basically understood by machines. Once you process your data in this format, you can leverage any kind of downstream Machine Learning or Deep Learning models based on the problem you're trying to solve. Real-world applications of Natural Language Processing include aspects like text classification, which you can use for example to find out whether an email is spam or not, classify news articles across different categories of news, group similar documents, and find out important entities from financial or healthcare documents. I am working with Constructor Learning to develop a solid NLP curriculum that focuses on understanding the fundamentals of how to process and load text data, how to clean text data, and then dive into areas of classical NLP. We look at aspects such as how to represent text data using classical NLP methods like words, n-grams, TFIDF and we also cover subjects like classification clustering, recommendation systems, search engines and so on. After this we also dive into Deep Learning for NLP looking at areas such as word embeddings, universal sentence embeddings as well as getting to know the newest area of Deep Transfer Learning for NLP by leveraging transformers which can be used to solve multiple tasks in NLP like sentiment analysis, question answering summarization, building search engines etc.”
What defines a Data Science influencer and why have you decided to associate with Constructor Learning?
“I think being a Data Science influencer is not just about sharing flashy content or trying to gain more views or likes but it is about having a genuine passion of sharing your knowledge with your community and to help people kick-start their own journey into Data Science. I have worked as a Data Scientist myself in diverse domains over the last several years like sales and marketing, supply chain, infrastructure, networking service management, and manufacturing semiconductors and I was even involved in product development for Artificial Intelligence-based products. There is a strong need for talented Data Scientists who can adapt to solving a wide variety of problems which can even be ambiguous, to begin with. I am a strong believer in quality over quantity. This is where I felt a strong synergy with the talented team at Constructor Learning who believe in being laser-focused on developing a solid curriculum and delivering it to students. Working with people and making an impact to help them start in the right direction and become better Data Scientists in the future inspires me.
Interested in reading more about Constructor Learning and tech related topics? Then check out our other blog posts.