Anvita Ravi

Assessed and benchmarked 5+ large language models, identifying nuanced behavioural differences in handling null values, ambiguous inputs, and contradictory prompts, contributing to improved model selection for production use. Led an in-depth analysis into the LLM-backend pipeline, mapping 5+ key AI-prompt interaction stages and enhancing the visibility and traceability of the full AI response lifecycle. Explored prompt token usage optimization by analysing internal log files, helping identify inefficiencies and proposing token reduction strategies.

Performed a meticulous examination of all power plants in Texas, categorizing them by size, age, and fuel type to assess their operational status and efficiency using Google Colab. Analysed energy production patterns for January 2023 and reviewed a 60-day report on bid prices for various resources, delivering critical insights into market trends and pricing strategies by employing Google Colab for data analysis. Conducted an in-depth report on Texas's energy sector, focusing on power production and resource procurement, providing a holistic understanding of the state's energy landscape.

Executed a stack buffer overflow exploit on a controlled server environment to simulate real-world threat scenarios for internal vulnerability assessments. Discovered and mitigated over 10 critical web application vulnerabilities including SQL injection, file upload flaws, and business logic errors by analysing PortSwigger labs using BurpSuite, supporting secure software development practices. Gained hands-on experience with proxy server configuration and ethical hacking techniques, enhancing the firm's ability to assess client systems and deliver informed security consultations.