arXiv subjects

Akhil Sharma

Publications and source records attributed to Akhil Sharma.

5 recordsLinked to original sources

arXiv2026

Retrieval-Augmented Reasoning for Chartered Accountancy

The inception of Large Language Models (LLMs) has catalyzed AI adoption in the finance sector, yet their reliability in complex, jurisdiction-specific tasks like Indian Chartered Accountancy (CA) remains limited. The models display difficulty in executing numerical tasks which require multiple steps while also needing advanced knowledge about legal regulations and the method of scaling their operations is not feasible in settings which have limited access to resources. We present CA-ThinkFlow as a parameter-efficient Retrieval-Augmented Generation (RAG) framework which operates with a 14B, 4-bit-quantized reasoning model, 14B-DeepSeek-R1, and a layout-aware Docling extraction system which maintains document structure during extraction. CA-ThinkFlow uses a basic RAG method which automatically adds retrieved information into the prompt, while it depends on the model's built-in Chain-of-Thought (CoT) functions to create context and produce correct answers. The system we developed system operates at performance levels which match large proprietary models when we tested it on the multi-level CA-Ben benchmark, achieving Scholastic Reliability Coefficient (SRC) results which equal 68.75\% of GPT-4o and Claude 3.5 Sonnet. The framework shows high efficiency and strength in handling parameters, but essential reasoning abilities fail to process complex regulatory texts which exist in fields such as Taxation.

cs.CL↗

arXiv2025

Bilevel Optimization for Covert Memory Tampering in Heterogeneous Multi-Agent Architectures (XAMT)

The increasing operational reliance on complex Multi-Agent Systems (MAS) across safety-critical domains necessitates rigorous adversarial robustness assessment. Modern MAS are inherently heterogeneous, integrating conventional Multi-Agent Reinforcement Learning (MARL) with emerging Large Language Model (LLM) agent architectures utilizing Retrieval-Augmented Generation (RAG). A critical shared vulnerability is reliance on centralized memory components: the shared Experience Replay (ER) buffer in MARL and the external Knowledge Base (K) in RAG agents. This paper proposes XAMT (Bilevel Optimization for Covert Memory Tampering in Heterogeneous Multi-Agent Architectures), a novel framework that formalizes attack generation as a bilevel optimization problem. The Upper Level minimizes perturbation magnitude (delta) to enforce covertness while maximizing system behavior divergence toward an adversary-defined target (Lower Level). We provide rigorous mathematical instantiations for CTDE MARL algorithms and RAG-based LLM agents, demonstrating that bilevel optimization uniquely crafts stealthy, minimal-perturbation poisons evading detection heuristics. Comprehensive experimental protocols utilize SMAC and SafeRAG benchmarks to quantify effectiveness at sub-percent poison rates (less than or equal to 1 percent in MARL, less than or equal to 0.1 percent in RAG). XAMT defines a new unified class of training-time threats essential for developing intrinsically secure MAS, with implications for trust, formal verification, and defensive strategies prioritizing intrinsic safety over perimeter-based detection.

cs.CR↗

arXiv2025

Indian Sign Language Detection for Real-Time Translation using Machine Learning

Gestural language is used by deaf & mute communities to communicate through hand gestures & body movements that rely on visual-spatial patterns known as sign languages. Sign languages, which rely on visual-spatial patterns of hand gestures & body movements, are the primary mode of communication for deaf & mute communities worldwide. Effective communication is fundamental to human interaction, yet individuals in these communities often face significant barriers due to a scarcity of skilled interpreters & accessible translation technologies. This research specifically addresses these challenges within the Indian context by focusing on Indian Sign Language (ISL). By leveraging machine learning, this study aims to bridge the critical communication gap for the deaf & hard-of-hearing population in India, where technological solutions for ISL are less developed compared to other global sign languages. We propose a robust, real-time ISL detection & translation system built upon a Convolutional Neural Network (CNN). Our model is trained on a comprehensive ISL dataset & demonstrates exceptional performance, achieving a classification accuracy of 99.95%. This high precision underscores the model's capability to discern the nuanced visual features of different signs. The system's effectiveness is rigorously evaluated using key performance metrics, including accuracy, F1 score, precision & recall, ensuring its reliability for real-world applications. For real-time implementation, the framework integrates MediaPipe for precise hand tracking & motion detection, enabling seamless translation of dynamic gestures. This paper provides a detailed account of the model's architecture, the data preprocessing pipeline & the classification methodology. The research elaborates the model architecture, preprocessing & classification methodologies for enhancing communication in deaf & mute communities.

cs.CV↗

arXiv2025

Large Language Models Acing Chartered Accountancy

Advanced intelligent systems, particularly Large Language Models (LLMs), are significantly reshaping financial practices through advancements in Natural Language Processing (NLP). However, the extent to which these models effectively capture and apply domain-specific financial knowledge remains uncertain. Addressing a critical gap in the expansive Indian financial context, this paper introduces CA-Ben, a Chartered Accountancy benchmark specifically designed to evaluate the financial, legal, and quantitative reasoning capabilities of LLMs. CA-Ben comprises structured question-answer datasets derived from the rigorous examinations conducted by the Institute of Chartered Accountants of India (ICAI), spanning foundational, intermediate, and advanced CA curriculum stages. Six prominent LLMs i.e. GPT 4o, LLAMA 3.3 70B, LLAMA 3.1 405B, MISTRAL Large, Claude 3.5 Sonnet, and Microsoft Phi 4 were evaluated using standardized protocols. Results indicate variations in performance, with Claude 3.5 Sonnet and GPT-4o outperforming others, especially in conceptual and legal reasoning. Notable challenges emerged in numerical computations and legal interpretations. The findings emphasize the strengths and limitations of current LLMs, suggesting future improvements through hybrid reasoning and retrieval-augmented generation methods, particularly for quantitative analysis and accurate legal interpretation.

cs.CL↗

arXiv2025

Lightweight Domain Adaptation of a Large Language Model for Legal Assistance in the Indian Context

In India, access to legal assistance for the general public has been observed to have a critical gap, as many citizens are not able to take full advantage of their legal rights due to limited access and awareness of apposite legal information. This paper thus introduces Legal Assist AI, a highly efficient framework designed to provide legal assistance in the Indian domain. The core contribution is a framework demonstrating how a smaller, 8-billion-parameter quantized model (Llama 3.1) can achieve superior domain-specific performance. This effective performance stems from integrating a Retrieval-Augmented Generation (RAG) system with strategic prompt engineering, supported by a high-quality, up to date corpus of more than 600 legal documents. This corpus includes the Indian Constitution and more importantly, the newly enacted Bharatiya Nyaya Sanhita (BNS) and Bharatiya Nagarik Suraksha Sanhita (BNSS) among others. Further, by achieving a score of 60.08\% in the All-India Bar Examination (AIBE) benchmark, the specialized approach based on RAG was found to be highly efficient and effective, improving on the 58.72\% score of the 175-billion parameter GPT-3.5 Turbo. It was also observed that the framework was able to manage and mitigate instances of hallucinations successfully, which is a critical requirement for practical legal applications. A Parameter Efficiency Index (PEI) is also introduced, with the goal of quantifying the superior efficiency that the framework was able to achieve, demonstrating how the 8B model is 22 times more parameter-efficient than the 175B baseline, and hence corroborating the potential of smaller domain-adapted models.

cs.CL↗