Perceptive Analytics

Software Engineer

May 2023 – Jun 2024

1. Worked with a client’s internal Risk team on providing explanations of actions (based on specified rules) taken by the platform on users such as banning them.

2. Populated a Hive table with contributing features which triggered the actions to be taken with a cronjob. Performed a progressive backwards search on these (action, feature) pairings to identify root causes with the help of an API call.

3. Leveraged AWS S3 for scalable data storage, and Apache Airflow to automate and schedule data collection workflows, resulting in a 35% improvement in data coverage of financial reports scraped from SEC and Bloomberg terminal.

4. Developed a semantic retrieval system using OpenAI embeddings leveraging Elasticsearch for scalable xxing and retrieval, reducing query latency and improving search extraction speed by 90%.

5. Containerised the Retrieval Augmented Generation(RAG) model using Docker and Amazon ECR and deployed as a serverless function in AWS Lambda.

Lakehead University, Canada

MITACS Research Fellow

May 2022 - August 2022

1. Selected for a fully funded onsite Research internship through the prestigious Globalink MITACS research internship program.

2. Developed a knowledge-grounded counter-narrative generation pipeline for online hate speech, leveraging large-scale pre-trained language models GPT-2 and XNLG for generating contextually relevant and informative responses.

3. Engineered a multi-stage knowledge retrieval and generation system utilizing transformer-based keyphrase extraction and BM25-based retrieval, integrating structured and unstructured external knowledge into counter-narratives.

Aided

Software Development Intern (part-time)

January 2021 – April 2021

1. Developed a multi-user, MERN stack healthcare application serving special needs children, parents, and therapists, featuring a synchronized scheduling and emailing system.

2. Utilized Docker for web app integration and deployed it via the NGINX local server.

Subtl.ai

Software Developer (part-time)

Jan 2023 - Mar 2023

1. Implemented a transformers based equivariance learning framework into a table-to-text generation CI-CD pipeline for ingesting tabular data.

2. Utilized RoBERTa, and T5 to enhance text retrieval and generation accuracy, resulting in a 26% increase in BLEU score on proprietary user data. New functionality included QnA over large Excel sheets and embdedded PDF tables.

3. I led the deployment of the RAG Tool at State Bank of India with 10000+ documents uploaded every month.

Precog-IIITH

Ungraduate Research Assistant

Jan 2022 - April 2023

1. Formalised and developed a novel noise-reduction framework using sentence-transformer LLMs. Buit a rumour-detection model ensembling LSTM and CNN models on top for improved detection of rumours from twitter. Paper under review at Expert Systems with Applications Journal.

2.Developed an MLP regressor model using m-BERT embeddings to evaluate the quality of synthetic machine-generated code-mixed sentences. Paper accepted at INLG 2022.

3. Built NER models for hashtag segmentation, scraped and processed 3.3M hashtags from Twitter using a loosely supervised approach. Paper accepted at LREC22.

Indian School of Business

Data Science Intern

Jan 2022 - April 2022

1. Developed an LSTM-based Seq-to-Seq activity forecasting system using Scikit-learn, PyTorch, on Netease Music users, processing over 56 million records from more than a million users. Achieved 79% accuracy.

2. Leveraged distributed computing with Apache Spark for scalable data preprocessing and implemented sliding window and seasonal decomposition techniques to enhance temporal pattern accuracy