Building A Meaningful Career In Ai
Meet Achintya Gupta
AI Engineer @ Nissan Motor Corporation
Achintya Gupta is an AI Engineer who firmly believe in artificial intelligence for social good. He writes building machine learning models, making sense and weaving a story from the data. He has experience in Deep Learning, Time Series forecasting and in making the business understand the intricacies of the models, enough to convince them to use it.
He likes trekking, writing @ag.ds.bubble and teaching.
How did you learn about artificial intelligence?
In quite an unconventional fashion, I would say.
During my college days, I was working on an IoT project to gather temperature and luminosity measurement of my room, and relay that to a cloud database, then transfer those measurements to an iOS Application which served as a controller to automate the control process to some extent.
I had to analyze the data that was gathered which then eventually led me to machine learning.
Following that, I did some online courses - Andrew NG’s being the first one of them, obviously and so on.
Then by the time I had to submit my Major project on Micro-Inverters, I was quite into Statistical Modelling and Simulations on MATLAB.
The learning process that started there has kept me going in this field, learning something new every day.
Eventually, I got my hands dirty with Deep Learning, which is an entirely different realm within machine learning.
How did you become an AI developer?
Pet projects, self-learning and getting myself some mentors.
I have to admit, for someone who had no formal training in Computer Science or Statistics, the process is a bit tedious, because these two fields make the bedrock on which the field of AI floats. But on the same time, if one attaches oneself to a mentor who can guide on these two fields, so much as to just point you in the right direction of which resource to read or how to take your project to next level that could translate into tremendous growth with a quick pace.
I was lucky enough to have some very smart brains alongside me throughout the journey, who helped me complete various pet projects of mine, something which I still keep on exploiting...
There is a plethora of information available online on machine learning and AI, but only seldom do they elucidate on what’s going behind the scenes. I think to become an AI developer, knowing how to use an algorithm won’t suffice one needs to have a fair understanding of the ingredients and recipe that went into making it, so a major chunk of my time is involved in research and implementation of it.
Also, an AI Engineer is expected to have some amount of Innovative or Creative mindset to think through the actual real-world process involved and be able to fit a model to it based on the findings, this process is quite taxing and something which I have usually seen lacking in AI Engineers.
Why are you interested in working with Deep Learning?
Quite frankly if you look at the basic neural network, its nothing but just matrix multiplication(some times a fancy multiplication) that goes through some activation functions in between and ends up mapping the data to the desired output and boom! you have yourself a Deep Learning model.
The idea that from the raw data, as simple as just pixel values or numerical encodings of text the model is able to adjust itself (train) to map the patterns within the data is very analogous to as to how a human brain works. That, right there is the most intriguing aspect of deep learning, for me at least.
Also if you look at the real-world applications of these models, I think advancements in medical sciences and solving social problems are where considerable ingenuity would be involved. I am in no capacity qualified to speak on these two subjects, but one day I would like to be.
Why Deep Learning and not another Ai subfield?
Well, my day to day work is a mixture of both, working with deep learning as well as other aspects of machine learning and statistical modeling.
As I pointed out earlier, statistics is one of the bedrock upon which AI floats, and statistical modeling is something which has been there for quite some time now and you will often find that old business usually having a well-integrated statistical process in place.
Which makes it a fairly decent weapon in your arsenal, but gradually we are seeing machine learning and AI being transcended into old business process, so going forth, the field of Deep Learning shows a lot of potentials.
Also, Statistical Modelling has been there for ages and it’s not going anywhere anytime soon, but Deep Learning is fairly new in comparison, which makes it all the more exploratory and exciting.
Will artificial intelligence take most of our jobs?
To some extent, it might, but I think it will augment into our lives rather than disrupting it or causing havoc.
Replacing a human being from the process has always been very hard if you just look at the invention of the computer and the effect it had, people thought that the more of it is available at a cheaper price the less of us would be required. But eventually, it just ended up at our workplace with us using them instead of us being ridden away by them.
Sure up-skilling or some training would be required at first, but we will soon adjust to it.