🚀 Exploring AI at Scale | Building Smarter Models | Advancing AI Research
I am a Machine Learning Engineer (MLE-II) at ServiceNow, specializing in AI, deep learning, and large-scale knowledge systems. I completed my MS-Research in Computer Science at IIIT Hyderabad, where my research focused on Geometric Deep Learning, Temporal Graph Neural Networks, and Large Language Models (LLMs). I specialize in federated learning, adversarial robustness, and scalable AI architectures, bridging cutting-edge research with real-world AI applications.
With 3+ years of experience in AI and data science, I have worked at Barclays as a Macro Structuring Analyst, where I developed AI-driven predictive models for financial markets, optimizing risk assessment and trading strategies for 10,000+ trades across 25 strategies. My expertise in LLM fine-tuning, AI model optimization, and knowledge graphs extends into both research and industry applications at ServiceNow.
My research work includes publications at top-tier AI venues including TMLR, AAAI, and workshops at KDD. I focus on federated learning, temporal graph neural networks, adversarial robustness, and optimizing LLMs for efficiency and security in real-world applications.
Introduces a novel framework combining spectral graph transformers with neural ordinary differential equations for federated learning on non-IID graph-structured data. Our approach leverages spectral graph transformations and continuous-depth modeling for robust feature aggregation while preserving global graph properties, enabling efficient cross-client knowledge sharing without direct data exchange. The framework demonstrates significant improvements in accuracy and convergence stability compared to standard federated GNN approaches, with applications in healthcare networks, financial risk modeling, and decentralized knowledge graphs.
Introduces an innovative low-resource adversarial framework for attacking Continuous Time Dynamic Graphs (CTDGs). Unlike resource-intensive traditional attacks, LoReTTA strategically identifies and poisons critical temporal edges with minimal computational overhead, exposing vulnerabilities in real-time graph-based AI systems.
By targeting temporal edge importance with limited perturbations, LoReTTA significantly degrades model performance while maintaining stealth. Our experiments across multiple real-world temporal graph datasets demonstrate substantial drops in predictive accuracy across diverse TGNN architectures, highlighting critical weaknesses in time-sensitive AI applications such as financial transaction monitoring, cybersecurity, and social network analysis.
Introduces a confidence-aware abstention mechanism for improving prediction reliability in Temporal Graph Neural Networks. By allowing the model to defer uncertain predictions, our approach enhances robustness in dynamic graphs that suffer from noisy or imbalanced data, critical for applications in fraud detection and financial modeling.
Abstention-based approach to reduce misclassification risk in Dynamic Graph Neural Networks through selective prediction strategies. Improves model reliability by prioritizing high-confidence predictions while avoiding misleading outputs in temporal graph applications.
Reliability-driven approach for Temporal Graph Neural Networks with confidence-based predictions. Demonstrates improved model trustworthiness and accuracy in low-data or high-noise scenarios across temporal graph benchmarks.
Combined Large Language Models (LLMs) with Graph Neural Networks (GNNs) to enhance reasoning over graph-structured data, introducing Subprocess and Island encodings. Subprocess encoding enabled directed graph reasoning, outperforming prior methods, while Island encoding improved global graph structure understanding by 25%. Conducted systematic evaluations across GPT-4, Claude, Llama, and Gemma LLMs, optimizing graph-to-text representation strategies for diverse reasoning tasks.
Leveraged Vision-Language Models (VLMs) as the backbone for zero-shot classification, extending Jiang et al.'s knowledge erasure algorithm to incorporate internal confidence scores. Our approach achieved robust classification performance on the CIFAR10 dataset by focusing on aggregating patch-level confidence scores. The research demonstrated improved robustness against adversarial settings, such as Gaussian noise and brightness variations, highlighting the potential of VLM-based confidence mechanisms.
💡 Machine Learning & AI: LLMs, NLP, Graph Neural Networks, Diffusion Models
🛠 Tech Stack: Python, PyTorch, TensorFlow, JAX, Hugging Face
🌍 Scalable AI Systems: Distributed ML, Federated Learning, MLOps
📊 Research & Development: Model Evaluation, Adversarial Robustness, Multimodal AI