Shakti Tiwari

I’m particularly interested in building AI systems that go beyond text generation and focus on decision-making, evaluation, and accountability. My interests include multi-agent systems, LLM judging and consensus mechanisms, safety gating, and auditability in AI workflows. I enjoy working on problems where models are required to justify outputs with evidence, assess risk, and support human judgment rather than replace it. I’m also keen on practical ML systems that are explainable, reproducible, and production-oriented rather than purely experimental.