Ashish Raj

Data Scientist

Bengaluru, Karnataka, India7 yrs 5 mos experience

Most Likely To SwitchHighly Stable

Key Highlights

Expert in building ML-driven products.
Strong background in data analysis and system design.
Published research on air-quality monitoring.

Stackforce AI infers this person is a Data Scientist with expertise in SaaS and research-driven analytics.

Contact

Skills

Core Skills

Machine LearningData Analysis

Other Skills

CClassificationCustomer AnalyticsCustomer LTV modellingData AggregationData VisualizationFeature StoreGeospatial AnalysisOla place tags and Ride UsecasesPythonResearchSQLStatistical ModellingTime-Series ForecastingUser Trustworthiness

About

I am a data-scientist interested in building ML driven products. Over the years, I have developed keen interest in whole ecosystem around data-driven products, from fundamentals of Statistics and Machine Learning to System-Design and Scalability of the solutions. Also, I do love to build utility tools for self and the team and firmly believe that putting efforts in that direction definitely pays-off in long run.

Experience

Walmart global tech india

Senior Data Scientist

Sep 2022 – Present · 3 yrs 6 mos · Bengaluru, Karnataka, India

Flipkart

Data Scientist

May 2021 – Jan 2022 · 8 mos · Bangalore Urban, Karnataka, India

Ola (ani technologies pvt. ltd)

2 roles

Data Scientist-II

Jul 2020 – Apr 2021 · 9 mos

· Feature Store: Wrote a utility Feature Store which comprises of a framework to define aggregation features across different data sources and a user facing API. API is written to be flexible with distinct time-frame for individual users and can aggregate features at mutliple levels: city, cluster of cities, country.
· User Trustworthiness: For user’s convenience of delayed payments, we built a model to learn a userbase who are trustworthy i.e. likely to pay within n-days of a rides. We identified a sizable chunk of the population with less than 0.5 percent chance of defaulting. A balanced bagging classifier was built on top of curated set of features for this skewed classification problem.
· Customer LTV modelling: We published a model of customer LTV prediction which is a corner-stone in customer life cycle management strategies at Ola. It is a time-series forecasting model for individual customer based on their previous interactions on the platform.
· Ola place tags and Ride Usecases: We have prepared a model to tag majority geohashes in a city with certain proprietary tags like transit, recreation etc.. It helps in understanding customer’s ride use-cases, which subsequently helps in other models like LTV.

Feature StoreUser TrustworthinessCustomer LTV modellingOla place tags and Ride UsecasesMachine LearningData Analysis

Research Engineer-I

Apr 2018 – Jun 2020 · 2 yrs 2 mos

· Ola Fleet as Dynamic Sensing Network: Solved the problem of identifying the optimal set of cars for the pilot of Air-Quality monitoring system in Delhi in collaboration with Microsoft Research. The work resulted in a paper Modulo: Drive-by Sensing at City-scale on the Cheap(https://dl.acm.org/doi/10.1145/3378393.3402275) adjudged as the best paper in ACM Compass’20.
· Counter-factual Modelling: Mentored an intern to build counter-factual model for hourly impact assessment on business metrics. Later as an extension of the project we also built a custom dashboard to visualise the impact and its statistical significance.