Inspection of telecom sites using mobile app which uses Image Processing and Computer Vision models to generate Inspection report for telecom sites. Reduced manual inspection efforts by 40%. Technology Used- Python, REST Api, Algortihm, Tensorflow, Object Detection.lg...Expand

Developed SaaS solution for autoscaling of worker nodes for V-Sphere cloud setup. Technology- Python, Elastic Search, Kibana, Shell Scripting, V-Sphere, Time Series Forecasting, Machine Learning, REST API.lg...Expand

Given a set of transactions, we need to identify which are recurring with a particular periodicity(e.g:- payroll transactions, power bill, Netflix subscriptions etc). This is being used as part of our financial wellness application to calculate the user's liability/income etc. Algorithm Used:- Siamese Network, LSTM, Graph Algorithm, Word Embeddings,Random Forest, Probabilistic Method, Sampling Algorithm. Technology Used:- Python, Tensof Flow, Keras, Scala, Akka, Java,lg...Expand

Created an automated framework using statistical methods for monitoring a ML-based application which deals with a high volume of data. This framework is responsible for monitoring the model after deployment in the production and generates alarm if the model is not performing as expected. Created Tableau Dashboard for visualization of model performance over the time. Algorithm Used:-Statistical Methods, Probability Distribution, Backtesting Technology Used:-SQL, Python, Tableau.lg...Expand

We are transforming the loan process for people who don't have credit history/credit score(FICO) yet but they have Ability to payback the loan amount. Algorithm Used:- Random Forest, XGBoost, Graph Algorithm Technology Used:- SQL, Java, Drools, Python, Data Structure.lg...Expand

Can we select a small subset of our data which can give the best financial market insight for the USA. We have a huge dataset, it is difficult to get the best insight from such a huge dataset because it consists of a lot of noise and requires huge processing power(directly proportional to cost). Algorithm Used:- Clustering Algorithm, Multiobjective Optimization, Genetic Algorithm. Technology Used:- R, Python, SQL, AWSlg...Expand

Can you predict a person's income class/ residential location/ work location/ employer/ family size using the aggregated bank and card transactions data. I was the sole owner of these project and currently it is consumed by multiple clients. Algorithm Used:- Majority Voting, Weighted Voting, DBSCAN Clustering, K-Means Clustering, XGBoost, Decision Trees Technology Used:- SQL, Apache Spark, Amazon EMR, R, Python, Tableaulg...Expand

Whenever a new version of a processor released, almost 70% of time spent on benchmarking, how can we optimize this effort using the state of the art machine learning technique, so that we can benchmark the newly made processor with much lesser time and very high confidence. Algorithm Used:- State of the Art classification and Regression Algorithm, Simulation, Advanced Graph Algorithm. Technology Used:- Java, Python, C, Architecture Simulation Tools SimpleScalar. lg...Expand

Detect political bias in the newspaper using natural language processing. We used Indian newspaper data as training and testing. Algorithm Used:- Standard NLP techniques(Tokenization, NER, Lemmatization, Stemming), RNN. Technology Used:- Java, Stanford NLTKlg...Expand