Rahul Dable

● Developing ML models with various algorithms and deep learning
techniques as per requirement, carry out feature engineering, perform
post training analysis and work on rebuilding retraining to improve model
performance until it meets the requirements, discuss results with product
managers.
● Periodically collect incremental data generated from various production
environments, retrain multiple BERT and custom trained W2V models
required in the main application on a regular basis.
● Constantly monitor performance of various ML models from production,
run the retraining cycle at specific intervals or whenever required on the
incremental data to maintain the performance..
● Suggest various approaches to automate the manual data tagging
process, try different techniques, models to annotate the data creation
process.
● Implemented various text summarization, question generation and
information retrieval and reader systems to support various functionalities
of the main application in the production.
● Attend daily scrum calls with product managers, testing, FE and BE
engineers to understand various AI-ML requirements, production bugs or
required changes in the main python application, address them
according to priorities.
● Participate in monthly sprint plannings, prepare detailed AI-ML job scopes,
planning all the activities according to their priorities and execution.
● Working with testing engineers to perform sanity on the latest changes in
python application and AI-ML models, deploying latest python changes
and AI-ML models from sprint cycles on productions.
● Try out various experiments with different NLP techniques, approaches,
algorithms etc, and suggest required changes in the current
implementation of the application if required.

Python Skills:
● Well conversant with Python Language and Python Libraries such as
Numpy, Pandas, Scikit-Learn, matplotlib, Regex etc.
● Performed Exploratory Analysis on various datasets.
● Manually implemented various machine learning algorithms using Python
such as TFIDF vectorizer, RandomSearchCV with k fold cross validation,
SGD Classifier with Logloss and L2 regularization, Computing Performance
metrics without Sklearn.
ML Algorithms:
● Studied various algorithms such as Logistic Regression, Linear Regression,
Naïve Bayes, Decision Trees, Random Forest, GBDT, XGBOOST etc and
applied them on various datasets such as Donor Choose Dataset.
● Built Recommendation Systems and Truncated SVD SGD algorithm to
predict ratings.
● Case study- Social network Graph Link Prediction (Facebook Friend
Recommendation)
● Built Microsoft Malware Classifier (Multi-Class Classification) using KNN, RF
and XGBOOST algorithms.
● ASHRAE Energy predictions case study.
Deep Learning:
● Familiar with the deep learning concepts of Neural Networks, MLP, CNN,
LSTM, and Transfer Learning.
● Studied text classification using CNN and built a Document Classification
system using CNN.
● CNN on CIFAR-10 dataset (DenseNet Architecture).
● LSTM on donor choose dataset.