I'm a
An aspiring AI researcher passionate about Explainable AI, AI Agents, and Multimodal AI. Currently exploring cutting-edge research while building practical solutions that bridge the gap between theoretical innovation and real-world impact.
I'm a third-year Computer Engineering student at DJSCE, Mumbai, specializing in AI/ML research with a particular focus on Explainable AI, AI Agents, and Multimodal AI. As Research Head at ACM DJSCE, I mentor students while pursuing my own research collaborations.
When I'm not diving deep into research papers or coding AI solutions, you'll find me cheering for FC Barcelona (Visça el Barça!) or writing on my blog to make ML concepts accessible to beginners. I believe in learning by doing and exploring diverse domains before specializing during my Master's studies.
Developing AI-agentic systems with dynamic graphs for factory operations, optimizing RAG-based legal document processing aligned with RERA rules, and evaluating computer vision models for retail environments including human detection and behavior analysis.
Designed MetaSearch, a search-augmented AI Agent for peer review consensus resolution. Developed novel LLM-argumentation systems evaluating 900+ ICLR reviews with high inter-annotator agreement (Cohen's Kappa = 0.834). Created heuristic frameworks based on Cognitive Load Theory for optimizing peer review information density.
Authored comprehensive whitepapers on LLM-based requirement generation, benchmarking leading models across diverse NLP metrics. Designed automated workflows in Azure AI Promptflow for RAG tasks, optimizing chunking strategies and reducing document retrieval perplexity by 15%.
Investigated C to Rust transpilation challenges, analyzing results from existing transpilers and CodeLLMs. Categorized bugs in LLM-generated code and improved code quality evaluation frameworks for future research applications.
Developed AI-driven marketing content pipelines by evaluating multiple image generation models. Created platform-specific content with customized AI-generated prompts and style guides for brand-aligned product advertisements using GenAI tools.
Multimodal, multilingual hate speech detection system with fine-tuned DistilRoBERTa, OCR-based text extraction, and unified video processing pipeline. Features AI Agent integration with comparative evaluation across multiple LLMs.
QR-based Police Feedback System empowering citizens with dynamic feedback reports and interactive geo-spatial visualizations. Features comprehensive NLP pipeline for multilingual support and sentiment analysis.
Computer Vision based healthcare solution for real-time asthma pump usage analysis. Implements advanced computer vision for posture detection, shake tracking, and pump-to-face distance measurement to provide actionable patient insights.
A novel AI Agent system combining disagreement detection and fact-grounded synthesis for automated meta-review decision-making in academic peer review processes.
A transparent framework for peer review quality assessment by combining LLM analysis with argumentation theory, achieving high agreement with human annotations in evaluating reviews.
A heuristic system to dynamically optimize the length of peer reviews by balancing brevity and thoroughness using information density and argumentation metrics to improve review quality.
An AI framework using LLM agents to automatically detect biases in peer reviews by analyzing sentiment, coherence, and consistency, offering actionable recommendations for conference organizers.
Comprehensive 70-page analysis benchmarking DeepSeek R1, GPT-4o, Gemini 2.0, and LLaMA 3.2 across diverse NLP metrics including BLEU, Levenshtein, and Jaccard Similarity.
Comprehensive study on multimodal, multilingual hate speech detection with practical real-world applications and impact assessment across various digital platforms.
Achieved All India Rank 499 in GATE Data Science and Artificial Intelligence examination among over 60,000 candidates, demonstrating strong foundational knowledge in AI/ML concepts.
Won the college competitive programming contest, showcasing algorithmic thinking and problem-solving skills essential for AI research and development.
Reached finals as ML Lead in the prestigious Rajasthan Police Hackathon 2024, developing innovative solutions for law enforcement challenges.
I'm always excited to discuss AI research, potential collaborations, or just chat about the latest developments in machine learning. Whether you're a fellow researcher, industry professional, or student, feel free to reach out!