Aviral Rastogi

Designed complex multi-tool prompts with context-aware task execution for autonomous AI agent systems. Built and evaluated AI agent workflows and evaluation frameworks to benchmark real-world problem-solving capabilities.

Developed probabilistic ML models (CNN, GAN, MLP) in Python to predict V50 ballistic limits of UHMWPE composite panels, analysing 300 experimental datapoints. Applied Probit regression and statistical analysis on ballistic test data to extract limit estimates with quantified uncertainty. Built LS-Dyna simulation pipeline to validate ML model outputs against finite element results ensuring physical

Led production unit of 40+ workers for luxury bath linen division; consistently achieved 95–98% production targets. Conducted data-driven production line studies to reduce defect rate and lower cost per kg of product.

Performed Pareto analysis on 5-year transformer failure data to identify root causes; proposed initiatives contributed to ~3% reduction in failure rate within 2 weeks. Reviewed and benchmarked quality standards, recommending process improvements across the transformer workshop.

Analysed Scope 3 value-chain reporting regulations and compared measurement methodologies across consumer product sectors. Researched and compared 50+ solution providers that offer supply chain transparency solution and emissions calculation.

Grade: 8.253/10

Activities and societies: Department Rank 3 out of 80+ students

Pursued a Bachelor of Technology in Production and Industrial Engineering at IIT Delhi, graduating in 2025 with a GPA of 8.253/10 and ranking 3rd in the department.