Dhamodhar Rathnam

Working as a machine learning engineer for the largest telecom provider to improve their manual tagging of their apps in their app store. Classifies apps into various categories based on their description. Performed classification with huge data and covered almost 50 plus languages in their app store and used CLD language detectors to detect the language. As a ML engineer, worked with various models testing it, logging it (logs at least 20 models per experiment), generating classification reports, and building training pipelines. All transformer models are built on top of pytorch. Worked with w2vec, ELMO, BERT embeddings. Worked with various ML and DL models like ECOC classifier, transformers like BERT and multi lingual models like XLM Roberta, LABSE, MBERT, Universal sentence encoder. Deploying the training pipeline and logging, monitoring the drift accuracy using MLOPS tools like mlflow, versionising with DVC. Explainable AI using SHAP, LIME tools. Performed Inference pipeline latency by using packages like Tensor RT, Onnx by quantizing or converting the models to other format.

Build predictive models using various machine learning tools to find out the SPAR for set of invoices. Develop algorithms using various Natural language processing and machine learning models for predictive maintenance. Esign algorithms to tag the SPAR to certain invoices using work bench for tagging invoices using manual research analysing the data for linked invoices and preparing reports using various analysis tools like excel. Deployed using jenkins using cognitive document processing to make onboarding processing easier for newly joined candidates using State of the art ML models. Evaluated diverse organizational systems to identify workflow, communication and resource utilization issues.

Built an email classification model using melusine a deep learning library, chatbot using IBM watson (testing only), solving adhoc tasks, and building predictive models to estimate customer score and classifying good customers. Performed adhoc data analysis using SQL.

Data analysis tasks using python building predictive models for smaller data. Manual analysis using excel. PI reports as per adhoc basis.