arXiv subjects

Raffi Khatchadourian

Publications and source records attributed to Raffi Khatchadourian.

13 recordsLinked to original sources

arXiv2026

DFAH-Bench: Benchmarking Observable Agent Instability in Financial Decision-Making

A financial AI agent can repeat a decision while changing the tools, order, or recorded arguments and results used to reach it. Outcome-only evaluation misses this variation, even when it matters for replay and change control. DFAH-Bench operationalizes the Determinism-Faithfulness Assurance Harness (DFAH), where faithfulness means fidelity of observable execution under replay, not answer correctness. The protocol qualifies comparable, sufficiently observed replays and measures decision agreement (DAR) and tool-path agreement (TAR) over the same eligible groups. We analyze 4,157 retrospective episodes from configurations with observed tool use across 719 synthetic compliance and financial DataOps groups, together with an argument-aware prospective extension comprising 570 eligible episodes across 190 groups. In that extension, decisions agree 94.2-95.1% while exact tool-name paths agree 66.9-69.4%, producing 25.8-27.3 percentage-point gaps; argument-and-result trajectory agreement falls to 45.0-51.5%. Even among unanimous-decision groups, paths vary in 66.7-68.9% under task weighting. DFAH-Bench makes the execution behind a stable decision visible for replay, investigation, and change review.

cs.AI↗

arXiv2026

Replayable Financial Agents: A Determinism-Faithfulness Assurance Harness for Tool-Using LLM Agents

LLM agents struggle with regulatory audit replay: when asked to reproduce a flagged transaction decision with identical inputs, many deployments fail to return consistent results. We introduce the Determinism-Faithfulness Assurance Harness (DFAH), a framework for measuring trajectory determinism, decision determinism, and evidence-conditioned faithfulness in tool-using agents deployed in financial services. Across 4,700+ agentic runs (7 models, 4 providers, 3 financial benchmarks with 50 cases each at T=0.0), we find that decision determinism and task accuracy are not detectably correlated (r = -0.11, 95% CI [-0.49, 0.31], p = 0.63, n = 21 configurations): models can be deterministic without being accurate, and accurate without being deterministic. Because neither metric predicts the other in our sample, both must be measured independently, which is precisely what DFAH provides. Small models (7-20B) achieve near-perfect determinism through rigid pattern matching at the cost of accuracy (20-42%), while frontier models show moderate determinism (50-96%) with variable accuracy. No model achieves both perfect determinism and high accuracy, supporting DFAH's multi-dimensional measurement approach. We provide three financial benchmarks (compliance triage, portfolio constraints, and DataOps exceptions; 50 cases each) together with an open-source stress-test harness. Across these benchmarks and DFAH evaluation settings, Tier 1 models with schema-first architectures achieved determinism levels consistent with audit replay requirements.

cs.AI↗

arXiv2026

Forecasting Future Language: Context Design for Mention Markets

Mention markets, a type of prediction market in which contracts resolve based on whether a specified keyword is mentioned during a future public event, require accurate probabilistic forecasts of keyword-mention outcomes. While recent work shows that large language models (LLMs) can generate forecasts competitive with human forecasters, it remains unclear how input context should be designed to support accurate prediction. In this paper, we study this question through experiments on earnings-call mention markets, which require forecasting whether a company will mention a specified keyword during its upcoming call. We run controlled comparisons varying (i) which contextual information is provided (news and/or prior earnings-call transcripts) and (ii) how \textit{market probability}, (i.e., prediction market contract price) is used. We introduce Market-Conditioned Prompting (MCP), which explicitly treats the market-implied probability as a prior and instructs the LLM to update this prior using textual evidence, rather than re-predicting the base rate from scratch. In our experiments, we find three insights: (1) richer context consistently improves forecasting performance; (2) market-conditioned prompting (MCP), which treats the market probability as a prior and updates it using textual evidence, yields better-calibrated forecasts; and (3) a mixture of the market probability and MCP (MixMCP) outperforms the market baseline. By dampening the LLM's posterior update with the market prior, MixMCP yields more robust predictions than either the market or the LLM alone.

q-fin.GN↗

arXiv2025

An Empirical Framework for Evaluating Semantic Preservation Using Hugging Face

As machine learning (ML) becomes an integral part of high-autonomy systems, it is critical to ensure the trustworthiness of learning-enabled software systems (LESS). Yet, the nondeterministic and run-time-defined semantics of ML complicate traditional software refactoring. We define semantic preservation in LESS as the property that optimizations of intelligent components do not alter the system's overall functional behavior. This paper introduces an empirical framework to evaluate semantic preservation in LESS by mining model evolution data from HuggingFace. We extract commit histories, $\textit{Model Cards}$, and performance metrics from a large number of models. To establish baselines, we conducted case studies in three domains, tracing performance changes across versions. Our analysis demonstrates how $\textit{semantic drift}$ can be detected via evaluation metrics across commits and reveals common refactoring patterns based on commit message analysis. Although API constraints limited the possibility of estimating a full-scale threshold, our pipeline offers a foundation for defining community-accepted boundaries for semantic preservation. Our contributions include: (1) a large-scale dataset of ML model evolution, curated from 1.7 million Hugging Face entries via a reproducible pipeline using the native HF hub API, (2) a practical pipeline for the evaluation of semantic preservation for a subset of 536 models and 4000+ metrics and (3) empirical case studies illustrating semantic drift in practice. Together, these contributions advance the foundations for more maintainable and trustworthy ML systems.

cs.SE↗

arXiv2025

LLM Output Drift: Cross-Provider Validation & Mitigation for Financial Workflows

Financial institutions deploy Large Language Models (LLMs) for reconciliations, regulatory reporting, and client communications, but nondeterministic outputs (output drift) undermine auditability and trust. We quantify drift across five model architectures (7B-120B parameters) on regulated financial tasks, revealing a stark inverse relationship: smaller models (Granite-3-8B, Qwen2.5-7B) achieve 100% output consistency at T=0.0, while GPT-OSS-120B exhibits only 12.5% consistency (95% CI: 3.5-36.0%) regardless of configuration (p<0.0001, Fisher's exact test). This finding challenges conventional assumptions that larger models are universally superior for production deployment. Our contributions include: (i) a finance-calibrated deterministic test harness combining greedy decoding (T=0.0), fixed seeds, and SEC 10-K structure-aware retrieval ordering; (ii) task-specific invariant checking for RAG, JSON, and SQL outputs using finance-calibrated materiality thresholds (plus or minus 5%) and SEC citation validation; (iii) a three-tier model classification system enabling risk-appropriate deployment decisions; and (iv) an audit-ready attestation system with dual-provider validation. We evaluated five models (Qwen2.5-7B via Ollama, Granite-3-8B via IBM watsonx.ai, Llama-3.3-70B, Mistral-Medium-2505, and GPT-OSS-120B) across three regulated financial tasks. Across 480 runs (n=16 per condition), structured tasks (SQL) remain stable even at T=0.2, while RAG tasks show drift (25-75%), revealing task-dependent sensitivity. Cross-provider validation confirms deterministic behavior transfers between local and cloud deployments. We map our framework to Financial Stability Board (FSB), Bank for International Settlements (BIS), and Commodity Futures Trading Commission (CFTC) requirements, demonstrating practical pathways for compliance-ready AI deployments.

cs.LG↗

arXiv2025

Speculative Automated Refactoring of Imperative Deep Learning Programs to Graph Execution

Efficiency is essential to support ever-growing datasets, especially for Deep Learning (DL) systems. DL frameworks have traditionally embraced deferred execution-style DL code -- supporting symbolic, graph-based Deep Neural Network (DNN) computation. While scalable, such development is error-prone, non-intuitive, and difficult to debug. Consequently, more natural, imperative DL frameworks encouraging eager execution have emerged but at the expense of run-time performance. Though hybrid approaches aim for the "best of both worlds," using them effectively requires subtle considerations. Our key insight is that, while DL programs typically execute sequentially, hybridizing imperative DL code resembles parallelizing sequential code in traditional systems. Inspired by this, we present an automated refactoring approach that assists developers in determining which otherwise eagerly-executed imperative DL functions could be effectively and efficiently executed as graphs. The approach features novel static imperative tensor and side-effect analyses for Python. Due to its inherent dynamism, analyzing Python may be unsound; however, the conservative approach leverages a speculative (keyword-based) analysis for resolving difficult cases that informs developers of any assumptions made. The approach is: (i) implemented as a plug-in to the PyDev Eclipse IDE that integrates the WALA Ariadne analysis framework and (ii) evaluated on nineteen DL projects consisting of 132 KLOC. The results show that 326 of 766 candidate functions (42.56%) were refactorable, and an average relative speedup of 2.16x on performance tests was observed with negligible differences in model accuracy. The results indicate that the approach is useful in optimizing imperative DL code to its full potential.

cs.SE↗

arXiv2023

Towards Safe Automated Refactoring of Imperative Deep Learning Programs to Graph Execution

Efficiency is essential to support responsiveness w.r.t. ever-growing datasets, especially for Deep Learning (DL) systems. DL frameworks have traditionally embraced deferred execution-style DL code -- supporting symbolic, graph-based Deep Neural Network (DNN) computation. While scalable, such development tends to produce code that is error-prone, non-intuitive, and difficult to debug. Consequently, more natural, less error-prone imperative DL frameworks encouraging eager execution have emerged at the expense of run-time performance. Though hybrid approaches aim for the "best of both worlds," using them effectively requires subtle considerations to make code amenable to safe, accurate, and efficient graph execution -- avoiding performance bottlenecks and semantically inequivalent results. We present our ongoing work on an automated refactoring approach that assists developers in specifying whether and how their otherwise eagerly-executed imperative DL code could be reliably and efficiently executed as graphs at run-time in a semantics-preserving fashion. The approach, based on a novel tensor analysis specifically for imperative DL code, consists of refactoring preconditions for automatically determining when it is safe and potentially advantageous to migrate imperative DL code to graph execution and modifying decorator parameters or eagerly executing code already running as graphs. The approach is being implemented as a PyDev Eclipse IDE plug-in and uses the WALA Ariadne analysis framework. We discuss our ongoing work towards optimizing imperative DL code to its full potential.

cs.SE↗

arXiv2022

QuerTCI: A Tool Integrating GitHub Issue Querying with Comment Classification

Empirical Software Engineering (ESE) researchers study (open-source) project issues and the comments and threads within to discover -- among others -- challenges developers face when incorporating new technologies, platforms, and programming language constructs. However, such threads accumulate, becoming unwieldy and hindering any insight researchers may gain. While existing approaches alleviate this burden by classifying issue thread comments, there is a gap between searching popular open-source software repositories (e.g., those on GitHub) for issues containing particular keywords and feeding the results into a classification model. This paper demonstrates a research infrastructure tool called QuerTCI that bridges this gap by integrating the GitHub issue comment search API with the classification models found in existing approaches. Using queries, ESE researchers can retrieve GitHub issues containing particular keywords, e.g., those related to a specific programming language construct, and, subsequently, classify the discussions occurring in those issues. We hope ESE researchers can use our tool to uncover challenges related to particular technologies using specific keywords through popular open-source repositories more seamlessly than previously possible. A tool demonstration video may be found at: https://youtu.be/fADKSxn0QUk.

cs.SE↗

arXiv2022

Challenges in Migrating Imperative Deep Learning Programs to Graph Execution: An Empirical Study

Efficiency is essential to support responsiveness w.r.t. ever-growing datasets, especially for Deep Learning (DL) systems. DL frameworks have traditionally embraced deferred execution-style DL code that supports symbolic, graph-based Deep Neural Network (DNN) computation. While scalable, such development tends to produce DL code that is error-prone, non-intuitive, and difficult to debug. Consequently, more natural, less error-prone imperative DL frameworks encouraging eager execution have emerged but at the expense of run-time performance. While hybrid approaches aim for the "best of both worlds," the challenges in applying them in the real world are largely unknown. We conduct a data-driven analysis of challenges -- and resultant bugs -- involved in writing reliable yet performant imperative DL code by studying 250 open-source projects, consisting of 19.7 MLOC, along with 470 and 446 manually examined code patches and bug reports, respectively. The results indicate that hybridization: (i) is prone to API misuse, (ii) can result in performance degradation -- the opposite of its intention, and (iii) has limited application due to execution mode incompatibility. We put forth several recommendations, best practices, and anti-patterns for effectively hybridizing imperative DL code, potentially benefiting DL practitioners, API designers, tool developers, and educators.

cs.SE↗

arXiv2022

A Tool for Rejuvenating Feature Logging Levels via Git Histories and Degree of Interest

Logging is a significant programming practice. Due to the highly transactional nature of modern software applications, massive amount of logs are generated every day, which may overwhelm developers. Logging information overload can be dangerous to software applications. Using log levels, developers can print the useful information while hiding the verbose logs during software runtime. As software evolves, the log levels of logging statements associated with the surrounding software feature implementation may also need to be altered. Maintaining log levels necessitates a significant amount of manual effort. In this paper, we demonstrate an automated approach that can rejuvenate feature log levels by matching the interest level of developers in the surrounding features. The approach is implemented as an open-source Eclipse plugin, using two external plug-ins (JGit and Mylyn). It was tested on 18 open-source Java projects consisting of ~3 million lines of code and ~4K log statements. Our tool successfully analyzes 99.22% of logging statements, increases log level distributions by ~20%, and increases the focus of logs in bug fix contexts ~83% of the time. For further details, interested readers can watch our demonstration video (https://www.youtube.com/watch?v=qIULoAXoDv4).

cs.SE↗

arXiv2021

Automated Evolution of Feature Logging Statement Levels Using Git Histories and Degree of Interest

Logging -- used for system events and security breaches to describe more informational yet essential aspects of software features -- is pervasive. Given the high transactionality of today's software, logging effectiveness can be reduced by information overload. Log levels help alleviate this problem by correlating a priority to logs that can be later filtered. As software evolves, however, levels of logs documenting surrounding feature implementations may also require modification as features once deemed important may have decreased in urgency and vice-versa. We present an automated approach that assists developers in evolving levels of such (feature) logs. The approach, based on mining Git histories and manipulating a degree of interest (DOI) model, transforms source code to revitalize feature log levels based on the "interestingness" of the surrounding code. Built upon JGit and Mylyn, the approach is implemented as an Eclipse IDE plug-in and evaluated on 18 Java projects with $\sim$3 million lines of code and $\sim$4K log statements. Our tool successfully analyzes 99.22% of logging statements, increases log level distributions by $\sim$20%, and increases the focus of logs in bug fix contexts $\sim$83% of the time. Moreover, pull (patch) requests were integrated into large and popular open-source projects. The results indicate that the approach is promising in assisting developers in evolving feature log levels.

cs.SE↗

arXiv2021

Interests, Difficulties, Sentiments, and Tool Usages of Concurrency Developers: A Large-Scale Study on Stack Overflow

Context: Software developers are increasingly facing the challenges of writing code that is not only concurrent but also correct. Objective: To help these developers, it is necessary to understand concurrency topics they are interested in, their difficulty in finding answers for questions in these topics, their sentiment for these topics, and how they use concurrency tools and techniques to guarantee correctness. Method: We conduct a large-scale study on the entirety of Stack Overflow to understand interests, difficulties, sentiment, and tool usages of concurrency developers. We discuss the implications of our findings for the practice, research, and education of concurrent software development, and investigate the relation of our findings with the findings of the previous work. Results: A few findings of our study are: (1) questions that concurrency developers ask can be grouped into a hierarchy with 27 concurrency topics under 8 major categories, (2) thread safety is among the most popular concurrency topics and client-server concurrency is among the least popular, (3) irreproducible behavior is among the most difficult topics and memory consistency is among the least difficult, (4) data scraping is among the most positive concurrency topics and irreproducible behavior is among the most negative, (5) root cause identification has the most number of questions for usage of data race tools and alternative use has the least. Conclusion: The results of our study can not only help concurrency developers but also concurrency educators and researchers to better decide where to focus their efforts, by trading off one concurrency topic against another.

cs.SE↗

arXiv2018

Proactive Empirical Assessment of New Language Feature Adoption via Automated Refactoring: The Case of Java 8 Default Methods

Programming languages and platforms improve over time, sometimes resulting in new language features that offer many benefits. However, despite these benefits, developers may not always be willing to adopt them in their projects for various reasons. In this paper, we describe an empirical study where we assess the adoption of a particular new language feature. Studying how developers use (or do not use) new language features is important in programming language research and engineering because it gives designers insight into the usability of the language to create meaning programs in that language. This knowledge, in turn, can drive future innovations in the area. Here, we explore Java 8 default methods, which allow interfaces to contain (instance) method implementations. Default methods can ease interface evolution, make certain ubiquitous design patterns redundant, and improve both modularity and maintainability. A focus of this work is to discover, through a scientific approach and a novel technique, situations where developers found these constructs useful and where they did not, and the reasons for each. Although several studies center around assessing new language features, to the best of our knowledge, this kind of construct has not been previously considered. Despite their benefits, we found that developers did not adopt default methods in all situations. Our study consisted of submitting pull requests introducing the language feature to 19 real-world, open source Java projects without altering original program semantics. This novel assessment technique is proactive in that the adoption was driven by an automatic refactoring approach rather than waiting for developers to discover and integrate the feature themselves. In this way, we set forth best practices and patterns of using the language feature effectively earlier rather than later and are able to possibly guide (near) future language evolution. We foresee this technique to be useful in assessing other new language features, design patterns, and other programming idioms.

cs.PL↗