Panitra Pandey

Built SayzoGuard, a multi-layered detection system preventing contact leakage, fraud, and policy violations in real-time communication. Developed LLM-based moderation pipelines with guardrails ensuring safe and deterministic production outputs. Designed multi-agent architectures for trust and safety workflows, improving modularity and traceability. Engineered scalable backend services with real-time inference integration using FastAPI. Implemented non-superficial health monitoring verifying true system dependencies for reliability. Improved detection accuracy and reduced false negatives in high-risk marketplace interactions.

Built an AI-powered coding assistant using NLP for real-time code suggestions and error detection. Developed a task distribution model using LLMs to optimize workload allocation and team efficiency. Enhanced an employee recommendation system by fine-tuning algorithms and integrating new data sources. Automated model deployment with CI/CD pipelines and containerized environments. Collaborated with cross-functional teams to deliver production-ready AI solutions.

Led development of an AI system automating Scope of Work (SOW) document creation using NLP and ML. Designed and trained models to interpret job descriptions and generate accurate SOWs. Integrated feedback loops to continuously improve model accuracy and adaptability. Reduced SOW creation time by 50%, improving efficiency and user satisfaction. Delivered the project end-to-end, from data preparation to deployment.

Robotics, Cultural and Sports