Karachi Institute of Economics and Technology

The purpose of the current novel study is to develop dual-coated multi-particulates with a combination of time-dependent inner and pH-dependent outer coating layers to control the release of the entrapped drug from the ascending colon onwards. Ibuprofen-loaded pellets prepared via powder layering technology were coated initially with time-dependent hydroxypropyl cellulose and ethyl cellulose-based inner polymeric layers and afterwards with the Eudragit L100 and Eudragit S100-based outer pH-dependent coating layers and evaluated. The best double-coated batch showed nominal in vitro release of 6.163 ± 0.23

In today's society with the increase in the population, there is an increase in unemployment as well. The gap between the candidates' knowledge and their performance in interviews is increasing. Students from well-funded institutions are trained well for their interview process and this gives them an edge to perform well in their interviews. Because the majority of students are enrolled in tier-2 and below institutes , they may lack this support from their colleges .To help solve this problem, we are proposing an AI-based mock Interview Generation platform which targets such candidates to assist them in their interview preparation. This platform uses AI to generate mock interviews based on the candidates' preparation needs and gives them a realistic interview environment. Our App generates realistic and domain specific interview questions. The results obtained from candidates using this app promises that we are able to achieve the promised outcomes and help them to identify , improve and strengthen their overall interview readiness. Our solution fills the gap present in the current scenarios by providing low-cost, available 24/7 interview sessions to help candidates prepare for their interviews

The modern power grid is rapidly transforming due to the increased integration of renewable energy sources (RESs) such as wind, solar, and plug-in electric vehicles (PEVs). The variable nature of RESs and the uncertain charging and discharging pattern of PEVs introduce significant uncertainty in power systems, making optimal power management more challenging. Therefore, it becomes necessary to develop robust and efficient strategies to ensure the economic and stable operation of power systems. Conventional techniques used for OPF often fails to capture these uncertainties, thereby motivating the use of metaheuristic algorithms for solving complex and non-convex OPF problems. This paper presents Aptenodytes Forsteri Optimization (AFO) algorithm, an optimization methodology inspired by emperor penguins’ foraging strategies for solving OPF problems. The algorithm is tested on two IEEE 30-bus configuration: (i) a conventional system having only thermal generators (ii) a modified system incorporated with RESs and PEVs with the main objective is to minimize the operating cost over a 24-hour scheduling horizon. Probability density functions are used to model the uncertainty related to wind speed, solar irradiance and PEV generation. The AGO-based OPF AGO demonstrate that there is significant enhancement in cost efficiency, reduction in peak operating costs, and system stability compared to conventional approaches. These findings established that AFO algorithm as an effective alternate tool for modern power systems management with high shares of renewable energy and electric vehicle integration.

Large Language Models continue to have a problem of factual accuracy in handling intricate structured texts. We introduce ParseRAG, an open-source, modular Retrieval-Augmented Generation pipeline which does not break down spatial and visual structure of documents during the knowledge extraction process. ParseRAG can maintain semantic coherence where other systems with text-only RAGs, based on fragmentation of multi-column layouts and disconnection between tables and their contextual descriptions, ParseRAG can use layout-aware parsing with PyMuPDF and LayoutParser with multimodal embeddings with LayoutLMv3. Our system uses hierarchical document representation which turns text and bounding box coordinates and visual features into dense vectors which are indexed using FAISS to allow targeted retrieval. ParseRAG makes high-quality document intelligence as a proprietary system available democratically to academic and enterprise users by offering a transparent, low cost option to traditional proprietary document processing systems. Codebase, model checkpoints and data on which evaluation is performed are publicly available so that reproducibility is possible.