Jan 2022 – Oct 2022 · 9 mos · Pune, Maharashtra, India

• Credit Risk Model: Built a high-precision Credit Risk model for credit card recommendations by leveraging financial and bureau data for 100M+ users. This improved credit risk assessments and approval rates.
• Data Automation Tool: Developed a PySpark-based automation tool for managing BI data requests, handling data of 100M+ users and 10M+ merchants, thus streamlining operations and reducing manual workload.

Apr 2020 – Dec 2021 · 1 yr 8 mos

• Built for fun and to solve a real problem, Notopass is an innovative Android app that allows users to share their Netflix, Amazon Prime, Spotify, Udemy, Coursera, and other digital accounts with friends, family, or colleagues—without ever exposing their actual credentials.
• Check out the App here - https://play.google.com/store/apps/details?id=com.notopass.app
• Achievements
• 750+ Upvotes and featured in weekly top products on Product Hunt- https://www.producthunt.com/posts/notopass
• 1500+ Downloads and 500+ active users.
• 20+ news coverage

Mar 2019 – Jan 2022 · 2 yrs 10 mos

• Open-sourced Chatbot specific Named entity Recognition, currently supports English and 6 most spoken Indic languages - https://github.com/hellohaptik/chatbot_ner
• Open-sourced Spell Correction Library Spello | Also available via pip https://github.com/hellohaptik/spello
• Created and presented models for text classification using Deep Learning Models to identify user intents in chat conversations. Achieved 30% better returns vs historical performance.
• Created chat analytics dashboard in R shiny dashboard for day to day analytics of conversations across platforms and able to figure out the main cause where bots fail to handle queries.
• Optimized the machine learning pipeline to reduce response times by 65 %.
• Created an App to showcase offline ML using Firebase ML Kit for Jio Set Top Box use case
• Active blogger for Haptik Tech - https://haptik.ai/tech/author/aman-srivastava/

Mar 2017 – Feb 2019 · 1 yr 11 mos

Sep 2016 – Feb 2017 · 5 mos · New Delhi Area, India

• Working as Data Scientist at Prophesee, Responsible for building new and additional features in the core product. Built a generalized Insights Engine based on NLP, Text Cleaning, Mining and Aggregations on top of ElasticSearch and Mongo. Major work areas include -
• Predictive models for social and digital engagements predictions.
• Recommendation engines for specific industries and specific channels.
• Forecasting and segmentation problems on social data
• Facebook Promoted Post Detection Problem using automated Machine Learning Models
• Identification and ranking Social Media Influencers, build a model to predict their Digital reach and
• build a model to estimating their Dollar Value for campaign based on their outreach, engagements,
• velocity and their followers growth.

Mar 2016 – Aug 2016 · 5 mos · Noida Area, India

• Working as data scientist at Gullak , a fintech data intelligence firm, working towards creating world class product in fintech Industry.
• Developing capabilities in analytics, recommendation engines, predictive modeling, machine learning and utilization of information as an asset via data processing, analysis, data mining, information trading and other means.
• Handling millions of rows of unstructured data and extracting insights using Natural Language Processing, Machine learning, Predictive modeling and Big data analytics.

Sep 2015 – Jul 2016 · 10 mos · Noida, Uttar Pradesh, India

• An online food ordering App for colleges and office. Schedule a time slot and get delivered on time. As students in colleges and employees work late at night, Foodgreed is for them, deliver healthy food at late night, so that you don't need to order junk food to fill up your stomach.

May 2015 – Jul 2015 · 2 mos · East of Kailash

• Worked on effectiveness of marketing campaigns by an F&B business by doing a qualitative and quantitative study of data provided
• 1. Exploratory data analysis on data provided of about 200 restuarents in 4 major cities. Identified
• major factors which affected the campaign
• 2. Visualised data using different libraries and packages in R like ggplot and charts
• 3. Build an Internal dashboard for day to day Monitoring.
• 4. Build a predictive Modeller to provide the best time and content for campaign lead generation.

Jan 2015 – Apr 2015 · 3 mos · Noida Area, India

• The project is about creating the dynamic dashboard of reports on a particular data that are sharable to others, the projects invoves uploading of sheets like excel and csv & data munging on these data and performing visualisation on them.
• Documentations - drive.google.com/file/d/0BhfdHhGgnUUYlNFeGllSEhJNE0/view1

Dec 2014 – Apr 2015 · 4 mos · Level 5, Tower C, 3C Green Boulevard Plot B-9/A, Sector-62, Noida 201307

• Worked on python Django web application for performing data pruning,analysis and reporting of raw data of students for schools
• Developed a web scraping tool in python for fetching schools data from different sources for acquisition
• Perform email marketing research to increase the sales of product.