Forecasting / Regression: Prototype, built and deployed XGBoost regression to predict the estimated time of arrival of a shipment for a logistics company. Airflow: Orchestration of AI pipelines using apache airflow on google cloud composer to preprocess, train and evaluate ML models. Classification: Designed and working upon an image classification model to automate the process of evaluation of shipments and avoiding human interference. Tools used: Google Cloud, Apache airflow, Google Data Studio, Google Bigquery, Docker, Google App Engine, Netezza, Scikit-learn, Tensorflow, Keras, Jira, Confluence, Github
TechnologiesGitDockerScikit-learnGoogle BigQueryDocker ComposeKerasTensorFlowGoogle App Engine
Recommender System: Prototyped, built and deployed a recommender system, using matrix factorization techniques, item based collaborative filtering and A/B testing, for a US client which helps their mobile users receive real time recommendations based on their past behaviour. Developed models using LightFM, implicit feedback, user-user collaborative filtering, Singular Value Decomposition Orchestration with apache airflow to automate the whole ai pipeline google cloud and big query for data analysis docker for containerisation of the model Tools used: Google Cloud, App Engine, Apache Airflow, google pub/sub, Asana, bitbucket, Google Bigquery, Cloud sql, Docker, RestAPI, Amplitude, Pandas, Numpy, Scikit-learn, Surprise, lightfm, Linux
TechnologiesPythonFlaskGitDockerGoogle CloudPub/SubREST APIsPandasRecommendation SystemsBigQuery
Teaching Data Science and machine learning to a full time batch of 35 students at Acamica institute of BsAs where students develop ML projects like 1. Image classifiers 2. Recommender system using Matrix Factorization and collaborative filtering 3. Natural Language processing 4. Using Watson API 5. Deploying trained models to the cloud
TechnologiesPythonETLLinear RegressionAzure Machine LearningNatural Language Processing (NLP)Data EngineeringText ClassificationRecommendation SystemsDeep LearningMachine Learning
Dashbo - A full stack application that helps marketing agencies to better manage the ad spending for their clients. Backend in Django + Postgres Frontend in ReactJS Cloud: AWS CI/CD with Gitlab Containerisations with Docker Task and queue processing with Celery and RabbitMQ Django, React.js, Postgres, Celery, AWS, CI/CD, Agile, Jira, Gitlab.
TechnologiesReactPythonDjangoAmazon Web Services (AWS)CeleryContinuous Delivery (CD)REST APIsRabbitMQBootstrapPostgreSQL
Edu Analytics - A full stack application that allows teachers to manage their students, make schedules, allot marks and send emails. Backend in Django + Postgres Frontend in ReactJS Cloud: AWS CI/CD with Gitlab Containerisations with Docker Task and queue processing with Celery and RabbitMQ Django, React.js, Postgres, Celery, AWS, CI/CD, Agile, Jira, Gitlab.
TechnologiesMySQLPythonDjangoGoogle CloudNumPyAWS Auto-scalingREST APIsPandasPostgreSQL
Maintained and improved Production support for a world class bank American Express (AMEX) while handling a team of 6 people. Provided production support from offshore Wrote bash scripts to automate tasks Support was conducted using an SMS ticketing system the three categories of severities The turn around time was Severity 1: 10 mins Severity 2: 4 Hours Severity 3: 24 Hours Developed and ran maintenance scripts on our servers since the system up time was 99.99% without compromise.
TechnologiesAgileData StructuresAlgorithmsDatabase PerformanceASP.NET CoreBash