At SOULAI, I architect and benchmark LLM training datasets, using multi-turn prompt design, reinforcement learning heuristics, and rigorous model response grading. My work has improved training dataset consistency and precision by 20%.
I evaluate code generation accuracy and complex algorithmic reasoning, and apply hallucination-reduction guidelines and safety alignment policies to prepare data for foundational AI model fine-tuning.
As a Cyber Security Analyst Intern at NOVANECTAR PVT. LTD., I engineered automated vulnerability scan workflows and alert filters that accelerated security triage by 35%. I also monitored system logs, analyzed anomalous network patterns, and documented incident mitigation playbooks.
In personal projects, I developed AI-powered healthcare and dynamic form platforms, along with a real-time weather dashboard. These projects use technologies including Next.js, Node.js, MongoDB, React.js, and REST APIs.
