Data scientists and machine learning engineers in India make about one-tenth of what their counterparts in the United States do, a leading global survey shows.
The median annual salary in India, based on 450 responses, is $11,715 (Rs 7.5 lakhs), a fraction of the comparable annual earnings in the US ($110,000). The median for all respondents from 52 countries, whose data was considered in the calculations, is $55,441.
Kaggle, the world’s largest global online community of data scientists, statisticians and machine learning engineers, published its The State of Data Science & Machine Learning annual survey earlier this week, deriving insights on 16,000 respondents in a report that polled the data science and machine learning industry.
The Google-owned platform currently boasts of over a million members and is known to attract the world’s smartest data scientists by holding public and private data science competitions. The findings from the survey provide key insights on demographics, salary, algorithms and tools used by the community.
What explains the yawning gap between the Indian data sciences community and the rest of the world?
Primarily, the proliferation of the term data scientist in India, which has become a catch-all term for anyone working with data, says Santanu Bhattacharya, a prominent Indian data scientist who has led teams at Facebook and logistics company Delhivery.
“If you look at Glassdoor, for the average data scientist salary in India, there are salaries being reported at Rs 3-4 lakh. These ‘data science’ jobs are offered by India’s IT service shops, which train fresh graduates who don’t necessarily have the background in math or statistics, and teach them some toolkits, without having an understanding of either the underlying business, or the underlying statistical models they are applying, leading to bad results,” says Bhattacharya.
It’s not an apples-to-apples comparison, he insists, comparing Indian talent to their western counterparts. “In the US, it is very defined. A data scientist is someone who has a strong math, science or computer science background from a strong school and can actually do their job from day one.”
This gulf in the talent stack is apparent in another chart, which compares the highest level of formal education among survey respondents. Bachelor’s degree holders make for more than half of India’s data science community on the Kaggle survey, with 33.8% holding a Master’s degree and just 3.8% PhDs. In the US, almost two-thirds of the respondents reported holding a Master’s degree (44.5%) or a PhD (20.7%).
Bhattacharya says that he receives around 20 resumes every day, out of which one or two candidates are a real representation of the profession. He thinks the overwhelming majority picks up cues from recruiters and resume parsing engines. “Here, [in India] anyone who is doing anything related to data is masquerading as a data scientist. Eventually what will happen is that there will be a massive correction and most will not be employable in real data sciences, but report generation, data quality control and all that wonderful stuff. They will be a peripheral workforce.”
Manish Saraswat, Category Head, Machine Learning at HackerEarth, a competitive coding platform, says that companies, especially in India, value proficiency over a Master’s or PhD degree, as they don’t have to spend much money on training the person.
“What they look for is basic level of education, a bachelor level degree in electronics or computer science. Companies are paying more attention to what kind of hands-on experience the candidate has in machine learning. That is where companies like HackerEarth and Kaggle come into the picture. On these open platforms, a BTech in civil engineering can show proficiency in solving business problems,” he says.
Kaggle’s survey finds that the median age for an Indian data scientist is 25 – one of the lowest in the survey and matched by the comparable age in countries such as Pakistan.
Saraswat attributes this to India’s demographic wave and a surge in interest in the profession, around the time it was called the ‘The Sexiest Job of the 21st Century‘ by Harvard Business Review. “The machine learning field has been in demand in India for the last 3-4 years, so people are still honing their skills to get into this field. The same trend would have started in US in 2010, when Kaggle was first founded. There’s a gap of a few years that separates the US and India,” he says.
Anshuman Das, Director, at recruitment process outsourcing firm CareerNet Technologies, says that the salary for a fresher data scientist at top companies such as Flipkart or Amazon is around Rs 18-20 lakh. “You don’t even get analytics talent for that,” he says, commenting on the salary figures reported by Kaggle. He also dismissed the average salaries for data scientists as reported on Glassdoor, as it uses consumer-fed data, which is not a reliable indicator. “A top-notch data science professional with five years of experience gets around Rs 40-50 lakh as compensation. It also depends on the company — a data science professional from Infosys will cost around Rs 15-20 lakh, while at Google India, it will cost Rs 50 lakh,” he says.
Startups that use machine learning and data science would pay around Rs 6-8 lakh for freshers, says Saraswat, while an IIT grad hired by an MNC bank would be on a package of Rs 15-20 lakh. Commenting on the low median salaries reported by Kaggle, he points out that at the 3-5 year experience bracket, the wage gap difference between the US ($95,000) and India ($18,744) is less stark.
Of the 2,700 Indian respondents in the Kaggle survey, 21.4% reported that they are not employed, but looking for work. Countries that report lower figures include here US (11.6%), Canada (12.3%), France (9.6%) UK (6.6%) and Israel (2.9). What explains this figure, and what does it say about the demand for data scientists in India?
“One thing I really feel is that there’s not enough opportunity. Many of us take up opportunities just for the sake of learning, and not for the salary,” says Simrat Hanspal, data scientist at Mad Street Den, a Chennai-based computer vision startup. Many of her peers had to do some internship to get some experience under their belt, before companies were willing to hire them, she says.
HackerEarth’s Saraswat is not sure if this data is indicative of a lack of demand. “I definitely believe there’s a skew in what people have answered or how they would have interpreted the question. People who would have answered that question would be college grads, who are in their final year,” he says.
Skilled data scientists are very much in demand, according to Das of CareerNet. “We are not getting enough of them. The top of the talent stack, which requires proficiency in machine learning or NLP (natural language processing), that population is very limited in India,” he says.