Prabhat Kumar

Experience

• 

  Tata Steel

  Jun 2014 - Jul 2014

  Research Intern

  Development of permanent magnet from steel industry by-products (Area: Hard Ferrites)Guide: Manish Kumar Bhadu, Prinicipal Researcher, Tata Steel- Selected for 6 week project to implement my idea through an open innovation challenge Mind Over Matter conducted by Tata Steel- Proposed and worked on the formation of ferrites by utilizing by-products such as slag and red oxide (alpha-Iron Oxide)- Prepared ferrites by mixing barium and strontium carbonate separately with iron oxide and doped them with slag in different proportions- Demonstrated that sample mixture with 10-20% of slag composition shows best magnetic properties through XRD and VSM experiments Show less

• 

  Mahindra Rise

  May 2015 - Jul 2015

  Intern

  Increasing productivity in S101 Engine assembly and testing through cycle time reduction (Area: Assembly Line, Cycle Time)Guide: Shiva Ryali, Manager Manufacturing, MHEPL• Optimization of operation cycle time in the engine assembly line using time-motion study and elimination of Non-Value added activities• Analysis of existing time required for different tasks at each workstation; identification of unnecessary operator movements• Implemented Line Balancing, Workstation Standardization (WSS), Kaizen Method; proposed to increase automation at some stations• Improved production efficiency of the assembly line through reduction in cycle time by around 16% Show less

• 

  CoinTribe Technologies

  May 2016 - Jul 2016

  Intern

  Guide: Rahul Kapil, Principal Data Scientist, CoinTribe Technologies Pvt. Ltd.Address Classification and Validation Algorithm (Area: Fraud Detection, NLP, Text Mining)- Developed a fraud detection algorithm to classify and give a confidence match score between declared and documented address- Pre-processed address strings; stripped prefix, suffix and other unwanted characters; used stemming and abbreviation correction- Extracted address components and classified them into different waterfalls; used Google Maps API to get GPS coordinates- Applied phonetics, spell correction, Levenshtein distance for string similarity percentage and weighed various scores- Published an API which returns final confidence match score and verdict as match or mismatch along with a valid standardized addressCompanies Ranking based on Google PageRank Algorithm (Area: Link Analysis, PCA)- Formulated probabilistic rank of companies based on inter-company transaction data over a certain time period- Used link analysis algorithm assuming debit and credit transactions as outgoing links and in-links respectively and amount as weights- Implemented regression model along with PCA over time-series datasets to predict companies' ranking in some future time Show less