Institution profile

Snowflake Inc.

Industry researchnorthamerica · us
Official website
Research library87linked papers
Opportunities87open roles
Selected work

Representative Papers

ConvCodeWorld: Benchmarking Conversational Code Generation in Reproducible Feedback Environments

Feb 27, 2025

Existing code generation benchmarks fail to model diverse multi-turn feedback—such as compilation errors, execution outcomes, and natural-language critiques—limiting rigorous evaluation of LLMs in conversational programming. Method: We introduce ConvCodeBench (static) and ConvCodeWorld (dynamic), the first reproducible multi-turn feedback evaluation benchmarks, establishing a feedback-driven evaluation paradigm. Leveraging GPT-4o, we generate structured natural-language feedback; integrate compiler-based error simulation; and deploy a coverage-aware execution engine to emulate realistic developer interactions. Benchmark consistency is validated via Spearman correlation. Contribution/Results: Experiments reveal that feedback type and intensity critically affect model adaptability: weaker models can surpass stronger ones’ single-turn performance after multiple feedback rounds, yet feedback-combination specificity induces generalization bottlenecks. Moreover, a trade-off exists between Mean Reciprocal Rank (MRR) and Recall, highlighting inherent limitations in current feedback-integration strategies.

1 citationsRead paper

Beware EviLLM: Enabling Vulnerability Injection via Large Language Models

Oct 02, 2026

This study addresses the realistic threat of third-party adversaries maliciously injecting security vulnerabilities through AI-assisted code generation pipelines. To this end, it proposes EviLLM, an attack framework that constructs a practical third-party threat model distinct from unintentional vulnerability introduction or insider threats. Specifically, EviLLM leverages compromised accounts or browser hijacking to intercept and manipulate both large language models and natural language specifications, enabling automated vulnerability injection via existing web-based attack vectors. Experimental results demonstrate that the proposed framework successfully introduces vulnerabilities spanning 13 Common Weakness Enumeration (CWE) categories. Furthermore, a user study confirms that most developers fail to detect these injected security risks. These findings highlight the significant fragility of the software supply chain within AI-assisted programming ecosystems.

0 citationsRead paper

Privacy-Friendly Cohort Determination: Sealed, CSP-Independent In-Browser ML Inference of Professional Segments for Identity-Less Advertising

Sep 28, 2026

This study addresses the challenge of identifying professional attributes for B2B advertising under third-party cookie restrictions and tightening privacy regulations. We propose SIF, a sealed inference framework that executes model inference client-side via WebAssembly, employs cross-origin iframes to bypass Content Security Policy constraints, and isolates network access through nested Workers to preserve user privacy. Furthermore, SIF integrates local differential privacy, randomized response, and k-anonymity mechanisms to generate privacy-preserving labels, which are seamlessly injected into the OpenRTB protocol for identifier-free precision targeting. Empirical evaluations demonstrate that the proposed approach effectively predicts company size and type while strictly bounding information leakage from adversarial models to five bits per site per week.

0 citationsRead paper
Recent publications

Latest Papers

Beware EviLLM: Enabling Vulnerability Injection via Large Language Models

Oct 02, 2026

This study addresses the realistic threat of third-party adversaries maliciously injecting security vulnerabilities through AI-assisted code generation pipelines. To this end, it proposes EviLLM, an attack framework that constructs a practical third-party threat model distinct from unintentional vulnerability introduction or insider threats. Specifically, EviLLM leverages compromised accounts or browser hijacking to intercept and manipulate both large language models and natural language specifications, enabling automated vulnerability injection via existing web-based attack vectors. Experimental results demonstrate that the proposed framework successfully introduces vulnerabilities spanning 13 Common Weakness Enumeration (CWE) categories. Furthermore, a user study confirms that most developers fail to detect these injected security risks. These findings highlight the significant fragility of the software supply chain within AI-assisted programming ecosystems.

0 citationsRead paper

Privacy-Friendly Cohort Determination: Sealed, CSP-Independent In-Browser ML Inference of Professional Segments for Identity-Less Advertising

Sep 28, 2026

This study addresses the challenge of identifying professional attributes for B2B advertising under third-party cookie restrictions and tightening privacy regulations. We propose SIF, a sealed inference framework that executes model inference client-side via WebAssembly, employs cross-origin iframes to bypass Content Security Policy constraints, and isolates network access through nested Workers to preserve user privacy. Furthermore, SIF integrates local differential privacy, randomized response, and k-anonymity mechanisms to generate privacy-preserving labels, which are seamlessly injected into the OpenRTB protocol for identifier-free precision targeting. Empirical evaluations demonstrate that the proposed approach effectively predicts company size and type while strictly bounding information leakage from adversarial models to five bits per site per week.

0 citationsRead paper