Apr 2021 – Feb 2022 · 10 mos

• Designed Data Science strategies and executed multiple projects for Naukri FastForward, E-Learning and Shiksha.
• Worked with top leadership to define high ROI Data Science projects with estimated impact on revenue growth of over 50 Cr annually.
• Designed end-to-end DS solutions for key recruitment based projects.
• Led a high performing team of Data Scientist on following projects
• o Lead Scoring model
• o Recruiter recommendation
• o Paid Job seeker recommendation
• o E-learning course recommendation
• o Paid job seeker ranking improvement in recruiter search
• Managed a team of 4 Data Scientists and 5 Interns.
• Established coordination between product, tech, operations, and my team for smooth and timely delivery of the Data Science projects.
• Ensured data accuracy and quantifiable output in structured deliverables for the team members.

Apr 2019 – Mar 2021 · 1 yr 11 mos

• Identified an opportunity in upgrading the search logic of recruiter search system by introducing personalization and reranking of results using vector space match. Convinced the business stakeholders to upgrade Elasticsearch from version 2.4 to version 6.4 and collaborated with multiple stakeholders to implement the new search logic. This increased NDCG of search results from 0.63 to 0.70.
• Was successful in establishing the inverse link between increased relevance and decreased usage for paid recruiter search system. The idea was accepted by sales team which formed a basis to increase the price of the product which resulted in significant increase in company’s revenue (~INR 50 crores).
• Collaborated with Operations team to build a recruiter spam detection system. The project involved guiding a junior data scientist to build a classifier that categorized queries into spam vs non-spam with an accuracy of 99.5%. About 2% of recruiters were flagged for spam and notified for taking top level action.
• Ensured optimum and stable utilization of server resources for different projects among multiple data scientists. We achieved 99.999% uptime for production APIs in search and recommendation projects.
• Incorporated personalization aspect in recruiter search autosuggestions leading to 60% increase in auto-suggestor usage. This enhanced recruiter’s search experience on our platform.
• Delivered a guest lectures:
• o MSRIT, Bengalure on “Design Thinking”
• o Deen Dayal Upadhyay Institute on “All about Data Science Projects”
• o Amity University on “Convolutional Neural Networks”
• Led a team of 4 data scientists on multiple projects: auto suggestors, recruiter spam detection, similar CV recommendation, and recruiter spam detection.

May 2015 – Mar 2019 · 3 yrs 10 mos

• Was responsible for the various data science projects that are powering up the naukri.com website. The details are:-
• Introduced content based Similar CV suggestion for over 12 million active candidates to increase quota consumption. Monetary impact is upwards of Rs 5 crore.
• Reduced the number of keystrokes to right suggestions by 25% using deep learning based Query Auto Completion. This immediately reduced recruiter's search effort.
• Addressed negative feedback behavior for over 80% of job seekers by identifying the negative aspects in the job such as unwanted location, experience range, salary range, skillsets and specific job requirements. The project increased the job relevance index by 10%.
• Achieved 25% improvement in search relevance by introducing state of the art search personalization in candidate search system (Resdex, www.naukri.com). Further, the project resulted in an increase of the number of resume downloaded per day thereby increasing monetization for Naukri by 10%. The key technologies involved are Elasticsearch, MongoDB and R which provides strong analytics capability and scalability.
• Developed content and search based similar CV suggestor for latest and low viewed CVs. This approach increased the coverage of similar CVS by 20% thereby increasing the number of CV views per day.
• Implemented CTC imputation data quality project where 10% of job seekers who mentioned their CTC as zero are replace with predicted CTC. This reduced spam mail and increased relevance in applies and Resdex search.
• Introduced 2000+ new upcoming skills and titles in the database and developed their associations with existing skills and titles using data mining and association rule mining systems.
• Implemented distributed computing for collaborative job seeker similarity to accommodate ever increasing number of job seekers (15 million plus).

Sep 2014 – May 2015 · 8 mos · Noida Area, India

• Served as Principal Data Scientist on a project of US based VoIP Company that provides telephone service via a broadband connection. The Company was monetarily losing in the order of million USD in their call center operations due to No Dial Tone (NDT) complaints. The objective of this project was to design and develop data science methodologies to crunch data from multiple data sources which amounted to over 1TB a day and obtain analytics driven actionable insights on an ongoing basis that would reduce the operational expenses, thus maximizing the return on investment (ROI) for the client’s investment on the project. Major tasks involved in the project were:
• Derive the relationship between network data and external weather data to establish the impact of weather on NDTs.
• Apply text mining to obtain relevant problem associate them with their actionable insights.
• Perform root cause analysis using robust statistical approach of single variable and multivariable analysis.
• Develop prediction models for predicting NDTs ahead of time using state of art prediction methodologies.
• The actionable insights obtained through my data science contribution in the project can be assessed from the fact that ROI is 6 months which amounts to significant year on year savings for the company.

Jan 2013 – Jan 2014 · 1 yr

Jan 2010 – Apr 2013 · 3 yrs 3 mos · Lausanne Area, Switzerland

• In addition to the research work, I am a system administrator of my lab and take care of about 30 linux workstations with central storage at a dedicated linux file server.

Jun 2009 – Sep 2013 · 4 yrs 3 mos · Lausanne Area, Switzerland

• I worked in the field of Non-invasive Brain Computer Interface (BCI) which deals with identifying signals from the brain recorded using Electroencephalogram (EEG). These signals are decoded using machine learning techniques to control external devices such as a wheelchair, robotic arm, on-screen speller. My work in particular involved the use of inverse solutions to identify the sources inside the brain that results in EEG on the scalp surface. From classification point of view, the features extracted from the estimated signal in intracranial sources might have better separability patterns and hence could increase classification accuracy of different mental commands. Thus the objective of my thesis was to effectively implement inverse solutions to increase the performance of BCI.

May 2008 – Jul 2008 · 2 mos · Lausanne Area, Switzerland

• Worked on project named "Polynomial Phase Signal (PPS)" which required development of a toolbox in MATLAB.
• It contained codes for 20 different algorithms (for both parametric and non-parametric methods) to handle higher order PPS with constant or varying amplitude.

Jun 2007 – Jul 2007 · 1 mo · Jamshedpur Area, India

• Successfully completed study on the calculation of mill motor parameters (current, power, torque) in hot strip mill set-up through process optimization models.