Rajiv Chaitanya

I’m a final-year Computer Science student at Dayananda Sagar College of Engineering with research interests in reinforcement learning, optimization, graph-structured learning, and quantitative finance. My work sits at the intersection of machine learning systems, financial modeling, and adaptive decision-making under uncertainty.

I’ve worked across AI research, econometrics, biomedical imaging, and enterprise LLM security. As a Research Intern at Indiana University Bloomington, I developed deep learning pipelines for ocular signal analysis, including preprocessing, feature extraction, and BiLSTM-based blink-phase segmentation from vertical EOG data. This work was later presented at ARVO Annual Meeting 2026.

Alongside this, I currently work as an Econometrics and Applied Research Intern at JSPM University, applying panel regression and time-series methods to macro-financial datasets using Python-based quantitative workflows. Previously, at SISA Information Security, I designed validation and safety pipelines for enterprise LLM systems involving intent classification, keyword filtering, sentiment-aware moderation, and adversarial robustness testing.

My research focuses heavily on reproducibility, deployment realism, and rigorous evaluation. I’ve authored work on reinforcement learning for portfolio optimization, resilience-aware allocation systems, swarm-augmented exploration for non-stationary RL, and streaming graph systems. My paper SRAS: A Lightweight Reinforcement Learning-based Document Selector for Edge-Native RAG Pipelines was selected as a Best Paper Candidate at ICEdge 2025, and my recent work includes research on graph stream partitioning, black-box trading strategy optimization, and regime-adaptive financial forecasting.

Beyond research, I also serve as a reviewer for IEEE conferences including IEEE World Congress on Computational Intelligence 2026 and IEEE International Conference on AI and Security for Industrial IoT Systems 2026.

I’m particularly interested in reinforcement learning for financial systems, adaptive graph intelligence, macro-financial risk modeling, and scalable ML systems that survive contact with reality instead of collapsing the moment the distribution shifts. The markets are chaotic, graphs drift, rewards lie, and that’s exactly what makes the problem beautiful.