arXiv subjects

Ishita Vohra

Publications and source records attributed to Ishita Vohra.

2 recordsLinked to original sources

arXiv2026

Towards Understanding, Analyzing, and Optimizing Agentic AI Execution: A CPU-Centric Perspective

Agentic AI serving converts monolithic LLM-based inference to autonomous problem-solvers that can plan, call tools, perform reasoning, and adapt on the fly. Due to diverse task execution need, such serving heavily rely on heterogeneous CPU-GPU systems with majority of the external tools responsible for agentic capability, either run on or are orchestrated by the CPU. Towards having a deeper understanding of its role, this paper aims to characterize and analyze the system bottlenecks introduced by agentic AI workloads from a largely overlooked CPU-centric perspective. We first present a compile-time characterization of agentic AI execution and choose representative workloads to capture the algorithmic diversity. We then perform runtime characterization of the representative workloads analyzing the end-to-end latency and throughput on two different hardware systems to isolate respective architectural bottlenecks. Based on the insights on the bottlenecks, we finally present two scheduling optimizations, namely, 1. CPU-Aware Overlapped Micro-Batching (COMB) and 2. Mixed Agentic Scheduling (MAS) on homogeneous and heterogeneous agentic workloads, respectively. In specific, these methods optimize for improved CPU-GPU concurrent utilization while reducing skewed resource allocation for heterogeneous execution. Experimental evaluations on the two hardware systems demonstrate the efficacy of COMB in yielding up to 1.7x lower P50 latency in standalone homogeneous workload execution and up to 3.9x/1.8x lower service/total latency under homogeneous open-loop load. Additionally, for heterogeneous open-loop load, MAS can reduce the total latency for minority request-type by up to 2.37x/2.49x at P50/P90 percentile.

cs.AI↗

arXiv2022

Is Twitter Enough? Investigating Situational Awareness in Social and Print Media during the Second COVID-19 Wave in India

The pandemic required efficient allocation of public resources and transforming existing ways of societal functions. To manage any crisis, governments and public health researchers exploit the information available to them in order to make informed decisions, also defined as situational awareness. Gathering situational awareness using social media has been functional to manage epidemics. Previous research focused on using discussions during periods of epidemic crises on social media platforms like Twitter, Reddit, or Facebook and developing NLP techniques to filter out relevant discussions from a huge corpus of messages and posts. Social media usage varies with internet penetration and other socioeconomic factors, which might induce disparity in analyzing discussions across different geographies. However, print media is a ubiquitous information source, irrespective of geography. Further, topics discussed in news articles are already newsworthy, while on social media newsworthiness is a product of techno-social processes. Developing this fundamental difference, we study Twitter data during the second wave in India focused on six high-population cities with varied macroeconomic factors. Through a mixture of qualitative and quantitative methods, we further analyze two Indian newspapers during the same period and compare topics from both Twitter and the newspapers to evaluate situational awareness around the second phase of COVID on each of these platforms. We conclude that factors like internet penetration and GDP in a specific city influence the discourse surrounding situational updates on social media. Thus, augmenting information from newspapers with information extracted from social media would provide a more comprehensive perspective in resource deficit cities.

cs.SI↗