Computational Social Science for a Trustworthy Information Ecosystem
I am a Master of Science in Computer Science graduate from USC and a computational social science researcher at USC ISI and UCLA. I build large-scale NLP and machine-learning systems to study how information — and misinformation — spreads across social platforms, from political discourse to online health misinformation. My work pairs production-grade data engineering with LLM-driven analysis to make digital information ecosystems more transparent and trustworthy.
Education
Master of Science in Computer Science
University of Southern California | Los Angeles, CA
Relevant Coursework: Natural Language Processing, Database Systems, Web Technologies, Data Structures & Algorithms, Multimedia.
USC's highest graduating distinction awarded for exemplary leadership and scholarship.
Bachelor of Engineering in Computer Science
NMAM Institute of Technology | Nitte, India
Relevant Coursework: Machine Learning, Object Oriented Programming, Computer Networks, Operating Systems, Compiler Design, UNIX.
Highest Honor for the College and the Computer Science Department for leadership and academic excellence.
Work & Research
Research Assistantships
- Health Misinformation NLP Pipeline: Engineered an automated pipeline (Python, OpenAI API) to classify large-scale TikTok datasets for a computational social science study of online health misinformation.
- Structured LLM Annotation: Built a zero-shot classification system with strict JSON-schema enforcement across a multi-dimensional academic taxonomy, then modeled the distribution of high-risk medical claims with cross-dimensional correlation analysis for academic publication.
- Large-Scale Election Discourse Analysis: Built end-to-end data pipelines (Python, Pandas, Apache Spark) to ingest and preprocess 50M+ TikTok/X posts, boosting throughput by 30%. Tuned BERTopic + UMAP workflows to collapse 700+ raw topics into coherent clusters, and authored sections of the public "2024 US Elections TikTok Multimodal Dataset."
- Harmful Content Detection: Designed computational taxonomies and trained ML classifiers (TensorFlow/PyTorch) to detect harmful online content at scale with 92% precision, integrating CI/CD workflows that achieved 95% test coverage.
- Assisted in managing the Database Systems course under Prof. Saty Ragavachari. Conducted office hours, crafted course materials, and addressed student inquiries. Served as Course Producer for the Database Systems course for two consecutive semesters.
Software Developer
Center for Homelessness Research, USC
- Led full-stack development of Android apps (Java/Kotlin) and responsive web portals (Angular), integrating secure REST APIs.
- Implemented GIS spatial analysis with ArcGIS SDK and GeoJSON pipelines to visualize and optimize datapoints on homelessness patterns.
- Automated data ingestion and batch processing using Python scripts, reducing manual ETL time by 60%.
Website Developer & Maintenance Engineer
Alphonso Consultancy | India
- Built and maintained ERP & website end-to-end: responsive UIs (HTML/CSS/JS), PHP/MySQL back-ends, and on-page SEO that boosted organic traffic.
- Prototyped blockchain and ML solutions: Ethereum/Hyperledger DApps for supply-chain tracking and Python-based chatbots sharpened smart-contract and data-science skills.
Publications & Patents
Tracking 2024 US Presidential Election Chatter on TikTok...
The Web Conference (WWW) 2025 | April 2025
Click to view publication
PatentE-Commerce Product Comparison System
Indian Patent Office | September 2024
Click to view document
Book ChapterPaddy Crop Disease Identification using Big Data ML
CRC Press | September 2021
Click to view chapter
Conference PaperDeep Learning Photograph Caption Generator
IEEE | Co-Author of Published Paper (November 01, 2021)
Click to view publication
PatentAn Automated Extraction and Summarization System
Indian Patent Office | September 2024
Brain Tumor Detection from MRI Images Using Deep Learning
3rd Int. Conf. on Data, Decision, and Systems | August 2024
Research & Development
Integrated DRAGON with LLaMA-7b-Chat
- Integrated DRAGON (Dense Retriever) with LLaMA-7b-Chat to reduce hallucinations in large language models.
- Employed dense retrieval techniques to ground model responses in accurate evidence from datasets like TriviaQA.
- Achieved up to 59% improvement in answer accuracy compared to baseline generation.
- View Project Document
Brain Tumor Detection from MRI Images
- Developed a CNN model for early detection of brain tumors from MRI scans.
- Implemented robust training pipelines to ensure high sensitivity and specificity in automated diagnostics.
- Achieved 89% test accuracy and 87.21% validation accuracy in real-world scan tests.
Deep Learning Photo Caption Generator
- Created a system using VGG-16 for visual feature extraction and LSTM for sequence prediction.
- Engineered the encoder-decoder architecture to translate visual semantics into natural language descriptions.
- Generated descriptive image captions with high BLEU scores across benchmark datasets.
- View Project Document
Detection & Visualization of Bone Abnormalities
- Executed an ML project using TensorFlow and CNN to identify irregularities on X-ray images.
- Implemented heatmaps and visualization tools to assist clinical professionals in abnormality identification.
- The Project was honored as the top in the department and received university funding for further exploration.
- View Project Document
Aspect-Based Sentiment Analysis
- Built a granular sentiment analysis engine to extract specific product features (aspects) from mobile reviews.
- Utilized NLTK and Scikit-learn to classify sentiments per feature, enabling detailed consumer feedback loops.
- View Project Analysis
Leadership & Involvement
Team Lead | Greater Los Angeles Homeless Count
Led the DemoGraphic Count initiative, coordinating large-scale data collection efforts across Los Angeles to support regional housing and health equity research.
NSS | Better India Mission
Implemented waste management solutions in villages and launched the 'Piggy Bank' project to support homeless individuals and people living with HIV.
Lead Organizer | HackRidea
Orchestrated a 24-hour National Level Hackathon. Managed logistics for 50+ teams and coordinated industry panels.
Tijuana, Mexico
November 2024: Collaborated with Esperanza to build houses for homeless people, providing hands-on construction and community support.
Colombia
March 2024: Engaged with indigenous Wayuu communities; assisted in building their first sustainable water tank for the village.
Skills & Technologies
AI & Data Intelligence
Engineering & Databases
Infrastructure, Cloud & Tools
Honors & Awards
National Overseas Scholarship
Awarded by the Karnataka State Government to support distinguished postgraduate research abroad (April 2024).
Representative of India
Represented the nation at the Asian Level YCS Conference in Indonesia (July 2019).
Top Data Collector
USC Suzanne Dworak-Peck School of Social Work (May 2023).
Best Project of University
NMAM Institute of Technology (March 2021).
Best Outgoing Student
Honored at both the College and CS Department level (April 2021).
Sakura Science Exchange
Selected by Japan Science & Technology Agency for research at Ritsumeikan University (2020).
Outstanding Achievement Award
Highest recognition for holistic excellence, academic rigor, and leadership at NMAMIT (April 2021).