Warning: Error while sending QUERY packet. PID=764179 in /home/wenger7/public_html/wewa/wp-includes/wp-db.php on line 1944
The human brain is a highly complex and spectacular gift of nature. Its capabilities are highly underrated. What is the difference between teaching a 6 year old the 10 numerals and an 18 year old a complex mathematical equation? It’s the nature of data, quantity of data and what sense the listener makes out of it. The human brain is designed to respond to, and understand data from patterns and repeated teaching. Imagine the same complexity to be expected from a machine. That’s where machine learning comes into picture. In basic terms, it is the use of a huge amount of data to identify patterns, formulate models and use them to “teach” a machine to solve a problem.
Today, any technology problem, which caters to a large scale of data needs analytics and machine learning as its backbone. The applications vary from healthcare imaging advances, complex numeric computations, web searches and recommendations, advertising re-targeting, optimization of logistics & fraud detection, to name a few.
A very common application is the recommendation system you notice while browsing through a store or relatable article. A purchase of a book from a series will recommend other books in the series whereas, the purchase of an electronic item like a microwave oven will recommend other accessories to use along with it, and not other ovens. This is the result of a well “trained” algorithm.
Primary players in search and ecommerce domains take pride in their machine learning expertise and their path breaking applications over the internet. A lot of startups and mid sized companies are now focusing on adapting and applying data analytics at the very beginning to prepare for the scale.
Data scientists, typically, come from a background in quantitative sciences like Statistics, Mathematics, Computer science or ideally an intersection of the three.
Here, you will find 4 key aspects to keep in mind, while searching for the ideal data scientist/ML scientist:
- Clash of the “Titles”: Often, there is a lot of confusion between the different categories in which these data players are classified. The below diagram gives the three basic steps in which a problem would be approached and also the skill set involved. Please note these titles, again, vary with companies. Also the diagram is applicable to an ecosystem of applied research only. The mathematical algorithms to be written utilising data and converting it to a format understood by a computer are the expertise of a machine learning scientist. The experts in Hadoop and BigData help in taking this code to the massive scale at which the system is operational flawlessly.
- Research v/s Application: Companies looking for data scientists often have some set of problems defined and are looking for experts who can translate this into an automated feature. On the other hand, some companies parallelly invest in researching new techniques and generating intellectual property in this space. This is a question often asked by a candidate especially if he/she holds a PhD. The patenting cultures of different companies vary based on business need and technology brand focus. A read through the candidate’s publications and patents gives a rough idea of their domain expertise-Speech recognition, NLP and quite recently deep learning and neural network applications for visual search.
- Corporate v/s Academics: A quick search for the data across the globe will show a good percentage of machine learning scientists pursuing the academic career path in esteemed universities like Stanford, CMU or University of Austin, Texas etc. In India they’re mostly with IISc, IIT KGP or IIT Delhi. A lot of industries fund good projects in such universities for the talent they posses. Also, corporates are open to working on fixed time assignments with them.
- Research and leadership: As focus on this technology increases, the need for leaders who are domain experts becomes key. This is a segment of people who are never away from technology. A successful leader in machine learning would be the one who can define a machine learning problem which, on solving, will translate to revenues and profits for the company. Building and running high performance ML teams would be an obvious addition. This would give rise to a lot of tech driven strategy development.
We, at Wenger & Watson Inc. are associated with internet, ecommerce and healthcare entities to identify machine learning and analytics talent for them.