AI Research Fellow & Tech Thinker
Exploring the intersections of machine intelligence, systems design, and the human experience. My research spans AI safety, emergent cognition, and adaptive computation.
Quantum computing sits at the frontier of computation itself — harnessing superposition, entanglement, and interference to tackle problems that remain intractable for classical machines. My research explores how quantum and AI can accelerate each other.
Variational quantum circuits (VQCs) as trainable layers — exploring how quantum kernels and hybrid quantum-classical architectures can outperform classical models on specific high-dimensional datasets.
QAOA and quantum annealing approaches to combinatorial problems — directly applicable to neural architecture search, hyperparameter tuning, and portfolio-style resource allocation in AI pipelines.
As quantum computers threaten classical cryptography, AI systems must evolve. Researching lattice-based and hash-based schemes to make model serving, federated learning, and data pipelines quantum-resistant.
Quantum simulators can model molecular interactions at native fidelity — unlocking drug discovery, materials science, and climate modeling at scales impossible for classical compute.
Near-term NISQ devices are noisy and limited. The practical path forward is tight classical-quantum co-design — where classical ML manages noise mitigation and quantum circuits handle exponentially hard subproblems.
DisCoCat and tensor-network approaches to grammar and meaning representation — early theoretical work suggesting quantum NLP could encode compositional semantics more naturally than classical embeddings.
"Quantum computing will not replace AI — it will give AI its hardest problems back, solved."
— Sai Sneha, Research Notes 2025Studying how LLMs acquire and represent knowledge, and what alignment means as models scale in capability and context.
Exploring architectures that allocate compute dynamically — letting models think harder on harder problems.
Understanding internal representations in transformer models to make AI systems safer, more transparent, and more predictable.
Building models that fluidly reason across text, vision, and structured data — closer to how humans integrate information.
Examining how AI systems affect society, fairness in machine learning pipelines, and paths toward responsible deployment.
Optimizing training infrastructure, efficient inference, and distributed learning to make research more reproducible and accessible.
Investigating quantum algorithms, variational circuits, and the potential of quantum advantage for machine learning tasks — from optimization to cryptography-resistant AI.
Studying the theoretical and practical pathways toward AGI — from reasoning and memory architectures to world models, agency, and the alignment challenges they introduce.
Open to research collaborations, speaking invitations, peer review exchanges, and thoughtful conversations about the future of AI.