Ali Nasir Bandukwala

Data Scientist

Physics Wallah•  June 2022 - Present

Project: Personalized Test-series using Adaptive Learning: User-Engagement through ‘Infinite Practice’ Impact: In The process of Launching personalized Ai based products by PW it is the first product to give personalized test questions to every user in ratios of Easy, medium and hard category depending upon his/her calculated knowledge using statistical modelling and engagement. oConceptualized state-of-art and innovative solutions, item response theory (IRT) to estimate the ability level of a person based on their performance on a test using difficulty, discrimination, speed and guessing parameters. oFormulated an algorithm from scratch for the 3Parameter Logistic model (a statistical model) to fit parameters with GPT-4 and optimized using Bayesian estimation to calculate knowledge parameters (θ) oCreated 2 APIs using FastAPI to return questions & optimized query to reduce latency from 40 secs to 530 ms. Impact: Automation of multi-class Classification from Manual tagging to cut down business cost and save hours. ▪ Project1- Automatic Tagging of Feedbacks by students into relevant issues: Pre-processed 818k Hinglish sentences; Vectorized using state-of-art LABSE Hinglish sentence transformers for optimal results; applied fine- tuning to XGBoost classification model to increase accuracy from 83 to 91% and achieved Recall value of 84%. ▪ Project2- Automatic bifurcation of SAARTHI Queries and Subjects: Performed EDA using classical NLP theory; Vectorized Hinglish sentences using TF-IDF for fast-processing; Comparative analysis between classical ML classification models; Altered threshold of Logistic regression to meet business demand of >85% Recall value. oCreated a cronjob-script from scratch and deployed through Airflow to daily update the classified tags. oSimplified Multi-class to binary classification to decrease latency and prepared classification reports.

Data Scientist

Vedantu•  June 2021 - June 2022

Project: VGyan (Doubt Solving through Ai) Impact: Built an automatic doubt solving chatbot for Ai Live sessions to help in solving academic/Non- academic doubts of the students on the platform itself without help of assistant teachers. oAutomated the process of classifying academic doubts & providing relevant answers for eachsubject and grade. oCreated sentence embeddings using distiIroberta and BERT- transformer and built a classifier with 92% accuracy & f1-score of 0.82, that can determine whether a doubt is academic or not and test its performance. oBuilt a regressor with r2 score 0.978 to predict the number of doubts asked in a respective session based on the attendance of students, so that the teachers can be allocated to the session for efficient resource utilization. Project: Post Session Comments Impact: Developed an ETL pipeline to get dispositions & sentiments of students’ comments to automate work of Students’ Account Manager in designating the students’ issues received after a session on platform oPerformed multi-class classification on the text-comments with an accuracy of 81%. oCreated and deployed a DAG in production using Apache-airflow that could: ▪ Automatically fetch real time data (session comments) from redshift database. ▪ Load final trained model pipelines and predict the multi-class categories for the comments.