cv

PhD in Computer Science

MS in Computer Science

BTech in Electronics and Communication Engineering

Graduate Teaching Assistant

• Assisted with and reviewed research proposals for the course Privacy in the Electronic Society

Data Scientist

• MLOps- As part of my MLOps work, I containerized applications & deployed them using GCP services & on Kubernetes clusters. Also used KServe (for model inferencing), MLflow (model experimentation and tracking), and Kubeflow pipelines for managing ML workflows.
• Entity Extraction- Used a multi-modal deep learning based architecture to extract data from unstructured pdfs.
• Constrained Optimization- worked on a constrained optimization problem using mixed integer programming.
• Retail Video Analytics- Transfer learned a Faster R-CNN based model for object detection. Implemented a deep learning based tracker.
• Conversational AI Chatbot- Built a chatbot using Rasa Machine Learning Framework and other external APIs (Google Maps API).
• PII Detection- Developed a Personally Identifiable Information Detection model by fine-tuning a pre-trained model for custom named entity recognition task. The application can be used to detect & mask sensitive data in text files, including files on GCP buckets. I used Apigee for API management and testing.
• Worked on a scheduling algorithm that matches members with medical service providers within various radii. I used multiprocessing which saved the firm 40 hours of execution time. I managed GCP compute instances, used cloud SQL for PostgreSQL, and used Plotly and Dash framework for developing an interactive web application.

Graduate Teaching Assistant

• I worked as a Graduate Teaching Assistant for the course Principles of International Business at W. P. Carey School of Business.
• As part of my work at the Behavioral Research Lab, I worked on an algorithm that categorized open-ended survey responses into two categories using NLP algorithms for topic extraction.

Research Intern - Software Engineering

• Worked on developing a new dimensionality reduction technique using tree-indexing.

Research Intern - Software Engineering

• Developed a Multi-Lingual Information Retrieval System that retrieves Wikipedia Articles across languages (French, Arabic and English).
• Used pre-trained Arabic, French and English word embeddings on monolingual vector spaces.
• Aligned these monolingual vector spaces using Orthogonal Procrustes Alignment Technique to get a multilingual vector space.
• Used t-SNE technique for dimensionality reduction and Bokeh for visualization purposes.

Research Intern - Software Engineering

• Analyzed different data sets to understand urban dynamics in fast-growing cities. Worked on a platform for integrating data from disparate physical and social sensors and implemented analytical approaches to mine such data and developed two applications as a result -- One, that visualizes various social demographics of Qatar and the other, an application called QEvents which visualizes various events happening in Qatar through a map-centered dashboard.

PhD Research

• Currently looking at using scalable LLMs to counter propaganda and disinformation online with a focus on analyzing linguistic techniques
• Research on detecting AI-generated content
• User Research using Quantitative Methods
• Fast Semantic Search using high dimensional vectors, Hadoop, Spark, and Milvus (a scalable vector database)

MS Research

• Continuous Glucose Monitoring & Prediction- Feature engineering of CGM time-series data. Train SVM and K-NN model to understand meal intake pattern. Forecasting of hypoglycemic events using RNN LSTM.
• Botnet detection on social media- CoorNet algorithm to detect coordinated inauthentic behavior.
• Pattern recognition- bayesian decision theory based classification use cases.

Undergraduate Research

• Project 1 - Built a face authenticated and voice-controlled robot that detects objects, and recognizes emotions from facial expressions using deep learning algorithms and models, on Raspberry Pi.
• Project 2 - Developed a Sign Language Recognition System (using two approaches — sensor-based and image-based)
• The sensor-based approach involves a glove that converts sign to speech using sensors such as flex sensors, contact sensors, accelerometers and gyroscopes, etc.
• The image-based approach was implemented with the help of Machine Learning and Computer Vision algorithms.

• GREPSEC VI Grant Recipient
• USENIX Security '23 Student Grant Recipient

• NYU School of Engineering Fellowship

• Machine Learning Engineering for Production (MLOps)

• Rasa Developer Certification

• AWS Fundamentals- Addressing Security Risk

• AWS Fundamentals- Going Cloud-Native

• Huawei Certified Network Associate

• Member of AI Ethics Committee - Atos zData

• Under-Secretary-General, Executive Board Affairs - NITD Model United Nations

• Human Rights Delegate - BITSMUN

• Student Volunteer - TEDxBirlaPublicSchool