Data Science Hiring Process at MobiKwik

Less than a week ago, Indian fintech unicorn MobiKwik emerged as the largest digital financial services platform for PPI Wallet transactions by value for April and May, as reported by the RBI, surpassing Paytm, PhonePe, and Amazon Pay, among others.

The IPO-bound company’s market share increased from 11% in March 2024 to 20% in April, and 23% in May of this year. MobiKwik has grown rapidly, with over 146 million registered users and 3.8 million merchant partners till now.

Founded almost 15 years ago by husband and wife duo Bipin Preet Singh and Upasana Taku, the company leverages AI and ML to solve complex business problems, improve its service offerings and make them personalised for customers.

Initially a mobile wallet provider, MobiKwik has expanded its offerings to include “buy-now-pay-later” credit, personal loans, merchant cash advances, wealth management, and insurance distribution.

“Our data science approach includes mining unstructured data (NLP), identifying patterns, and deriving meaningful insights to assist businesses in making better decisions,” Saurabh Dwivedi, SVP of technology, MobiKwik.

The company is currently expanding its workforce and is looking for skilled data scientists across different domains.

Inside MobiKwik’s Data Science Ops

According to Dwivedi, one of the primary challenges that the Peak XV backed startup’s data science team addresses is credit risk assessment, which holds tremendous importance for any fintech company.

It has developed various ML models for products such as MobiKwik ZIP, EMI, and Xtra to enhance the decision-making process and improve the quality of the loan portfolio.

The ZIP underwriting model, for example, is a proprietary ML model that assesses the creditworthiness of users applying for ZIP. It integrates traditional (bureau) and alternative data (device), along with behavioural data from in-house wallets, to forecast the likelihood of users becoming delinquent.

This model, built using data from over a million approved ZIP users, outputs a probability score, categorised into deciles from one (lowest risk) to 10 (highest risk).

Similarly, the Spend Analyser tool categorises unstructured data using regular expressions, pattern identification, keywords, and historical trends. It identifies savings opportunities and explores investment options such as MobiKwik’s Xtra 14%, fixed deposits, mutual funds, and lending products.

Additionally, the team employs various data extraction techniques to parse documents and draw insights from diverse structured and unstructured datasets. Along with this, the company is powering its chatbot with LLMs to respond to users’ transactional queries in a personalised way.

Tech Stack

MobiKwik’s data science projects leverage a diverse tech stack, including SQL, Python, PySpark, PyTorch, TensorFlow, Keras, LLMs, REST APIs, Docker, and GitLab CI/CD. These tools and frameworks are selected based on the specific requirements of each project to build robust and scalable data science solutions.

Moreover, the team utilises GPT and Transformer Models for NLP tasks, such as transaction pattern analysis, entity extraction, and document analysis.

Interview Process

“We evaluate candidates on a comprehensive set of skills, including expertise in machine learning models, deep learning, NLP, conversational AI, and generative AI skills,” Dwidedi told AIM, stating that the selection process also considers multimodal communication, creativity, curiosity, and resilience.

Proficiency in prompt engineering, CI/CD, model monitoring, and key performance indicators (KPIs) are essential criteria for candidates.

The interview process at MobiKwik consists of several stages:

  1. Screening Round: Candidates interact with the HR and the hiring team to discuss their professional journey and work experience.
  2. First Technical Round: This round assesses the candidate’s coding practices, NLP proficiency, and ability to implement ML and DL solutions.
  3. Second Technical Round: A comprehensive evaluation of technical skills, soft skills, and industry experience.
  4. Managerial Discussion: Focuses on overall fitment and the candidate’s impact.

HR Fitment Round: An overall evaluation of cultural alignment with the organisation’s ethos.

Expectations

Candidates joining MobiKwik’s tech team can expect to work on innovative products driven by data science, such as risk modelling, UPI wallet, Lens, ZIP, ZIP EMI, and Xtra.

“Our large data platforms are constructed with extensive data pipelines based on machine learning. We are focused on real-time use cases, using multi-faceted AI capabilities. This serves as a great opportunity for candidates to grow into full-stack ML engineers and data scientists, thereby adding significant value to our team,” he added.

On the other hand, the company expects candidates to contribute to business scaling, demonstrate expertise in relevant technologies, and commit to continuous learning.

Work Culture

MobiKwik aims to foster a dynamic work culture that includes direct interaction with founders and the senior leadership team, ensuring maximum facetime. “With a grounded leadership ethos valuing attitude over aptitude, we maintain a lean and flat hierarchy,” explained Dwivedi.

The company also provides special perks for employees, including group health insurance, personal accident insurance, and special leave allowances. Additionally, employees can take a recharge leave of five consecutive days after completing three years of service, promoting work-life balance and personal rejuvenation.

Check out MobiKwik’s careers page here or drop a mail at ta@mobikwik.com.

The post Data Science Hiring Process at MobiKwik appeared first on Analytics India Magazine.

Follow us on Twitter, Facebook
0 0 votes
Article Rating
Subscribe
Notify of
guest
0 comments
Inline Feedbacks
View all comments

Latest stories

You might also like...