Valentin Malykh

Frequent revisions of complex regulatory documents in large organizations often introduce inconsistencies and contradictions that are difficult for lawyers and auditors to detect manually. Existing tools rely on character-level diffs and therefore miss paraphrases and semantic shifts. We introduc...

Retrieval-augmented code generation commonly conditions a decoder on large retrieved snippets, which couples online cost to repository size and introduces long-context noise. We present Hierarchical Embedding Fusion (HEF), a two-stage repository representation for code completion: (i) an offline...

Frequent revisions of complex, lengthy regulatory documents in large organizations often create inconsistencies and contradictions that lawyers and auditors find difficult to detect manually. Existing tools rely on character-level diffs that fail to capture paraphrases and semantic shifts. We pre...

Modern code agents achieve strong results on challenging software engineering benchmarks such as SWE-bench, but solving each issue remains expensive: most inference cost is spent not on patch generation, but on repository exploration and search, accounting for up to 56% of tokens in our experimen...