Sep 2019 – Apr 2022 · 2 yrs 7 mos · Hyderabad, Telangana, India

Mar 2018 – Sep 2019 · 1 yr 6 mos · Gurgaon, India

• Acquired by Naspers/PayU.
• PaySense is a fin-tech startup providing access to personal and consumer credit in India. It has constantly appeared amongst top lending fin-tech startups in India.
• Lead a team of seven (Scientists and Engineers), responsible for developing and maintaining Underwriting, Risk and Collections models
• Developed a Domain Specific Language (Canvas) for document parsing., document here refers to any rendered target like a PDF, Word file, Image, Screen shot etc. The Document processing service based on Canvas has resulted in 20% reduction in operating costs by eliminating the reliance on third party services and human auditors.
• Developed a free from document scanner for mobiles for high quality document image acquisition
• Developed an alerting system to respond and report any anomalous trends in user acquisition and loan disbursals
• Developed signature matching and document verification systems

Feb 2017 – Mar 2018 · 1 yr 1 mo · Bengaluru Area, India

• MileIQ is a automatic mileage tracking app developed by a startup in SFO which was acquired by Microsoft
• Developed high performance image classifier for mobile devices with processing of < 10ms per image., this model is part of MileIQ app and helps in detecting the images of receipts and invoices
• Developed MileIQ User Profiling service using Spark, ADLS and CosmosDB

Feb 2016 – Feb 2017 · 1 yr · Bengaluru Area, India

• Lead of team of 3 developers to migrate MileIQ backend (services, databases, caches, CI/CD etc) from AWS to Microsoft Azure stack
• Worked on MileIQ integration with Microsoft Outlook Calendar

Jun 2014 – Feb 2016 · 1 yr 8 mos · Mumbai Area, India

• Housing.com is a listings portal for residential real estate. Was the first to provide a usable map based listings search experience
• Developed Multiverse Recommender System using Tensor Factorisation, which out performed the existing model by 2000 bps of Mean Reciprocal Rank (MRR) in email recommendations and 1800 bps of MRR on live website
• Built Hybrid Recommender System using Collaboration via Intent paradigm, which combines the merits of Content Based and Collaborative Filtering methods., it also helps in tackling the cold start problem associated with the ephemeral inventory and users of the Housing.com (or any real estate portal for that matter) with elegance. This model outperformed the existing model (content based) by 5000 bps
• Developed User Preference Modelling using FP-Growth approach replacing Vector Space Model in order to capture, constraints and conditional preferences of the users
• Developed ETL system which serves master data to all the data products, Business Intelligence and Product Analytics groups of Housing.com

Jan 2014 – Jun 2014 · 5 mos · Bengaluru Area, India

• Worked on backend services of Consumer Engagement Platform
• Developed Entity Resolution service on SAP HANA stack