Data Scientist (Advanced Analytics)

My responsibility as a Lead Engineer here in Lrnr, brought opportunity to look at the data coming to the server at high speed, in various format, and mainly unstructured. The real time recommendation for "next" activity for the learner became crucial for us. Scenario got complicated because we expect approx one million users hit the server for various reasons at a given moment, and each learner is unqiue with its behaviour and learning state, we needed Big Data analysis. Currently we are using Big Data solutions, with inhouse development processes. I am the part of the team with the responsibility to setup, look after and guide the process so that our goal is achieved and maintinaed flawlessly.

I am the team-lead and working on adaptive e-learning algorithms for our flagship product "lrnr". Lrnr as we call it, is a user centric online tutoring model, where every learning activity of the learner is captured. We use data-mining methodologies to categorise the data across the users and across learning and assessment contents. We also look at specific learning style of the user, and the specific teaching style of the instructor. We use statistical methods such as Bayesian Network, Item Response Theory and Cognition models to derive and callibrate user's effective model. These models are dynamic in the sense that it depends on current learning state of the specific user. Suggestions and learning content reorganization for that user is done using these effective models. I work on conceptualization, modeling and calibration of the effective models.

"Digiage ( http://www.facebook.com/diigitalage?fref=ts) " is a start-up in the field of social media and online reputation management. I worked on creating a computational platform to analyze informations gathered from social media nodes, such as facebook, twitter, and other review and blog pages. This tool has helped them in understanding the correlation of concept building, blogs and reviews and linking it to individual and group behaviour. These analysis has them in creating effective strategy for their clients. The tool is good for identifying, creating and calibrating mental models of the bloggers so that one can effectively choose blogging content for Reputation Risk Management.

I was one of the co-founders of Finquark Technologies Pvt Ltd, Mumbai, India. Our aim was to produce low-cost low resourse consuming software solutions, consultancy and training for the business requirements related to finance domain. We had developed Credit Risk and Market Risk Calculator according to Basal-2 conventions and Reporting tool, and Value at Risk (VaR) calculation and reporting tool with has robust methods for scenario analysis, backtesting and if-else analysis. The design of the software caters to a large spectrum of users. For beginners it is point and click, while for advanced user, these can be used as an analysis tool. We used various advanced statistical algorithms such as Kalman filters, regression analysis based on Garch variants, stochastic differential equations for interest rate and volatility modeling etc.

I was involved with various training programs conducted by Aptivaa Consulting for their clients. I have also worked on derivative pricing and validation project for Aptivaa.

My job requirements was to design and implement and test Interest Rate Models and Volatility models for the pricing of various exotic derivatives and structure. I have also contributed to deal-pricing activity of Volframe Analytics Pvt Ltd Mumbai.

My responsibilities include: Developing algorithm for derivative pricing, market risk, value at risk and credit risk computation.