{"id":"6f5c0cc70e5464280e79d201c9836a63e82516879f72ea26b6bacc61123af871","title":"arXiv cs.AI paper says review and deployment constrained enterprise code migrations","argument":"Mario Corchero and coauthors report in an arXiv paper that Automated Uplift used LLM-assisted pipelines to migrate code across thousands of Bloomberg repositories. The multi-month initiative covered two campaigns: Python packaging and the C++ build system. Together, they generated 2,333 pull requests and resulted in 729 successful migrations.\n\nThe authors’ central finding is that producing proposed code changes was easier than getting them reviewed, tested, and deployed by the teams that own the repositories. Their hybrid architecture maximized deterministic processing, confined the LLM to bounded file transformations within a pre-computed context, and incorporated production telemetry such as execution logs and usage traces. The authors say costs for owning teams fell when changes could be verified in the pipeline before opening a pull request, as with C++ compilation and test validation, and when repositories were uniform enough for fixes to carry over.\n\nThe authors present these experiences as patterns for architecting hybrid pipelines and scaling code migrations. Their abstract reports outcomes from two campaigns; it does not establish how the approach would perform elsewhere.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The evidence is the arXiv abstract; it does not provide detailed campaign methods or results beyond the summary given there. The reported outcomes come from two Bloomberg campaigns and do not establish how the patterns would perform in other organizations or migrations.","speaker":{"name":"Corchero, Mario, Williams, David, Kollar, Laszlo Kiss, Liu, Haowen, Bairoliya, Shivam, Malbasic, Gala, Kirbas, Serkan, Sarro, Federica","roleAtSource":"Article author"},"source":{"title":"Hybrid LLM-Deterministic Pipelines for Enterprise-Scale Code Migration at Bloomberg","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2611.05457","publishedAt":"2026-11-03","dateBasis":"Date basis not provided."},"evidence":[{"text":"The paper reports two Automated Uplift campaigns, for Python packaging and a C++ build system, producing 2,333 pull requests and 729 successful migrations in total.","excerpt":"Together, these campaigns generated 2,333 pull requests, resulting in 729 successful migrations.","url":"https://arxiv.org/abs/2611.05457","relationship":"Source-stated"},{"text":"The authors identify review, testing, and deployment by repository-owning teams—not generating migrations—as the bottleneck.","excerpt":"generating migrations turned out to be the easier part; the bottleneck was the cost, for owning teams, of reviewing, testing, and deploying them.","url":"https://arxiv.org/abs/2611.05457","relationship":"Source-stated"},{"text":"The authors say costs fell when changes could be verified in the pipeline before a pull request and when fixes could carry over among sufficiently uniform repositories.","excerpt":"These costs were lowered when changes could be verified directly in the pipeline before opening a PR (as with C++ compilation and test validation), and when the population was uniform enough for fixes to carry over from one repository to the next.","url":"https://arxiv.org/abs/2611.05457","relationship":"Source-stated"},{"text":"The authors describe an architecture that favors deterministic processing and restricts the LLM to bounded file transformations using pre-computed context.","excerpt":"maximizes deterministic processing, confines the LLM to bounded file transformations within a pre-computed context","url":"https://arxiv.org/abs/2611.05457","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.05457","checkedAt":"2026-11-10T01:56:07.104Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How well would the authors’ patterns transfer to migrations beyond these two campaigns?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"arXiv cs.AI paper says review and deployment constrained enterprise code migrations","summary":"Mario Corchero and coauthors report in an arXiv paper that Automated Uplift used LLM-assisted pipelines to migrate code across thousands of Bloomberg repositories. The multi-month initiative covered two campaigns: Python packaging and the C++ build system. Together, they generated 2,333 pull requests and resulted in 729 successful migrations.\n\nThe authors’ central finding is that producing proposed code changes was easier than getting them reviewed, tested, and deployed by the teams that own the repositories. Their hybrid architecture maximized deterministic processing, confined the LLM to bounded file transformations within a pre-computed context, and incorporated production telemetry such as execution logs and usage traces. The authors say costs for owning teams fell when changes could be verified in the pipeline before opening a pull request, as with C++ compilation and test validation, and when repositories were uniform enough for fixes to carry over.\n\nThe authors present these experiences as patterns for architecting hybrid pipelines and scaling code migrations. Their abstract reports outcomes from two campaigns; it does not establish how the approach would perform elsewhere.","question":"How well would the authors’ patterns transfer to migrations beyond these two campaigns?","claims":[{"text":"The paper reports two Automated Uplift campaigns, for Python packaging and a C++ build system, producing 2,333 pull requests and 729 successful migrations in total.","excerpt":"Together, these campaigns generated 2,333 pull requests, resulting in 729 successful migrations.","support":"source-stated"},{"text":"The authors identify review, testing, and deployment by repository-owning teams—not generating migrations—as the bottleneck.","excerpt":"generating migrations turned out to be the easier part; the bottleneck was the cost, for owning teams, of reviewing, testing, and deploying them.","support":"source-stated"},{"text":"The authors say costs fell when changes could be verified in the pipeline before a pull request and when fixes could carry over among sufficiently uniform repositories.","excerpt":"These costs were lowered when changes could be verified directly in the pipeline before opening a PR (as with C++ compilation and test validation), and when the population was uniform enough for fixes to carry over from one repository to the next.","support":"source-stated"},{"text":"The authors describe an architecture that favors deterministic processing and restricts the LLM to bounded file transformations using pre-computed context.","excerpt":"maximizes deterministic processing, confines the LLM to bounded file transformations within a pre-computed context","support":"source-stated"}],"limitations":["The evidence is the arXiv abstract; it does not provide detailed campaign methods or results beyond the summary given there.","The reported outcomes come from two Bloomberg campaigns and do not establish how the patterns would perform in other organizations or migrations."],"mediaId":"8f4b0d0d78c3519749ff9a76c8dbbe9da0b66d1e4e1f2012890c6bca26adee0a","correctionReason":null,"id":"6f5c0cc70e5464280e79d201c9836a63e82516879f72ea26b6bacc61123af871","publicationId":"0074f38d-5a09-4682-858a-3f59511ab19e","version":1,"sourceUrl":"https://arxiv.org/abs/2611.05457","sourceTitle":"Hybrid LLM-Deterministic Pipelines for Enterprise-Scale Code Migration at Bloomberg","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-11-03","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-05T05:06:33.248Z","createdAt":"2026-11-10T01:25:10.085Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"8f4b0d0d78c3519749ff9a76c8dbbe9da0b66d1e4e1f2012890c6bca26adee0a","kind":"article","url":"https://arxiv.org/abs/2611.05457","sourceUrl":"https://arxiv.org/abs/2611.05457","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-10T01:24:18.362Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.05457","checkedAt":"2026-11-10T01:56:07.104Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"8f4b0d0d78c3519749ff9a76c8dbbe9da0b66d1e4e1f2012890c6bca26adee0a","confidence":0.81,"provider":"typesafe","model":"jev-1.13.0","inputHash":"57e21a7e8bac1bf11d557f55355e7aff3f271c3930615d094ced37ad62fc1411","decidedAt":"2026-11-10T01:24:20.353Z"},"provenance":{"textHash":"7dfc14df7f48f71caa4383dba4132f7c2b3511204ef79f8940013c16c8bcf08e","observationId":"a1996bdda4a9750ac5ae1d1884a8737b84309c472e9abb8c988e234a062dd478","retrievedAt":"2026-11-10T01:24:18.043Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"c6ee7c60-885d-4703-8d4b-ffd12c392631"},"reviewer":{"model":"gpt-6-luna","runId":"51a4fe9a-d35a-4834-8769-d6f08a9671a0"}},"audit":{"status":"passed","reviewedAt":"2026-11-10T01:25:10.085Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Corchero, Mario","url":null,"kind":"unknown"},{"name":"Williams, David","url":null,"kind":"unknown"},{"name":"Kollar, Laszlo Kiss","url":null,"kind":"unknown"},{"name":"Liu, Haowen","url":null,"kind":"unknown"},{"name":"Bairoliya, Shivam","url":null,"kind":"unknown"},{"name":"Malbasic, Gala","url":null,"kind":"unknown"},{"name":"Kirbas, Serkan","url":null,"kind":"unknown"},{"name":"Sarro, Federica","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Corchero, Mario","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Corchero, Mario","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05457","locator":"head/meta[name=author|citation_author]"},"assertionId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T01:56:07.104Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f","sourceUrl":"https://arxiv.org/abs/2611.05457","contentSha256":"d1c7572c9f145afb701333a91f30c32a8275ce7483243f6de546dd68ef9a0df4","textHash":"7dfc14df7f48f71caa4383dba4132f7c2b3511204ef79f8940013c16c8bcf08e","observedAt":"2026-11-10T01:56:07.104Z"}},{"actor":{"name":"Williams, David","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Williams, David","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05457","locator":"head/meta[name=author|citation_author]"},"assertionId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T01:56:07.104Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f","sourceUrl":"https://arxiv.org/abs/2611.05457","contentSha256":"d1c7572c9f145afb701333a91f30c32a8275ce7483243f6de546dd68ef9a0df4","textHash":"7dfc14df7f48f71caa4383dba4132f7c2b3511204ef79f8940013c16c8bcf08e","observedAt":"2026-11-10T01:56:07.104Z"}},{"actor":{"name":"Kollar, Laszlo Kiss","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Kollar, Laszlo Kiss","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05457","locator":"head/meta[name=author|citation_author]"},"assertionId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T01:56:07.104Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f","sourceUrl":"https://arxiv.org/abs/2611.05457","contentSha256":"d1c7572c9f145afb701333a91f30c32a8275ce7483243f6de546dd68ef9a0df4","textHash":"7dfc14df7f48f71caa4383dba4132f7c2b3511204ef79f8940013c16c8bcf08e","observedAt":"2026-11-10T01:56:07.104Z"}},{"actor":{"name":"Liu, Haowen","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Liu, Haowen","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05457","locator":"head/meta[name=author|citation_author]"},"assertionId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T01:56:07.104Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f","sourceUrl":"https://arxiv.org/abs/2611.05457","contentSha256":"d1c7572c9f145afb701333a91f30c32a8275ce7483243f6de546dd68ef9a0df4","textHash":"7dfc14df7f48f71caa4383dba4132f7c2b3511204ef79f8940013c16c8bcf08e","observedAt":"2026-11-10T01:56:07.104Z"}},{"actor":{"name":"Bairoliya, Shivam","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Bairoliya, Shivam","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05457","locator":"head/meta[name=author|citation_author]"},"assertionId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T01:56:07.104Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f","sourceUrl":"https://arxiv.org/abs/2611.05457","contentSha256":"d1c7572c9f145afb701333a91f30c32a8275ce7483243f6de546dd68ef9a0df4","textHash":"7dfc14df7f48f71caa4383dba4132f7c2b3511204ef79f8940013c16c8bcf08e","observedAt":"2026-11-10T01:56:07.104Z"}},{"actor":{"name":"Malbasic, Gala","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Malbasic, Gala","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05457","locator":"head/meta[name=author|citation_author]"},"assertionId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f:participation:5","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T01:56:07.104Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f","sourceUrl":"https://arxiv.org/abs/2611.05457","contentSha256":"d1c7572c9f145afb701333a91f30c32a8275ce7483243f6de546dd68ef9a0df4","textHash":"7dfc14df7f48f71caa4383dba4132f7c2b3511204ef79f8940013c16c8bcf08e","observedAt":"2026-11-10T01:56:07.104Z"}},{"actor":{"name":"Kirbas, Serkan","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Kirbas, Serkan","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05457","locator":"head/meta[name=author|citation_author]"},"assertionId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f:participation:6","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T01:56:07.104Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f","sourceUrl":"https://arxiv.org/abs/2611.05457","contentSha256":"d1c7572c9f145afb701333a91f30c32a8275ce7483243f6de546dd68ef9a0df4","textHash":"7dfc14df7f48f71caa4383dba4132f7c2b3511204ef79f8940013c16c8bcf08e","observedAt":"2026-11-10T01:56:07.104Z"}},{"actor":{"name":"Sarro, Federica","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Sarro, Federica","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05457","locator":"head/meta[name=author|citation_author]"},"assertionId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f:participation:7","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T01:56:07.104Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f","sourceUrl":"https://arxiv.org/abs/2611.05457","contentSha256":"d1c7572c9f145afb701333a91f30c32a8275ce7483243f6de546dd68ef9a0df4","textHash":"7dfc14df7f48f71caa4383dba4132f7c2b3511204ef79f8940013c16c8bcf08e","observedAt":"2026-11-10T01:56:07.104Z"}}],"participationCoverage":{"total":8,"returned":8,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2611.05457","snapshotId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f","contentSha256":"d1c7572c9f145afb701333a91f30c32a8275ce7483243f6de546dd68ef9a0df4","textHash":"7dfc14df7f48f71caa4383dba4132f7c2b3511204ef79f8940013c16c8bcf08e","observedAt":"2026-11-10T01:56:07.104Z","basis":"publisher-page-metadata","sourceLinks":45,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/6f5c0cc70e5464280e79d201c9836a63e82516879f72ea26b6bacc61123af871/source"}},"canonicalUrl":"https://escapemeridian.com/readings/arxiv-cs-ai-paper-says-review-and-deployment-constrained-6f5c0cc7","formats":{"html":"/readings/arxiv-cs-ai-paper-says-review-and-deployment-constrained-6f5c0cc7","markdown":"/readings/arxiv-cs-ai-paper-says-review-and-deployment-constrained-6f5c0cc7.md","json":"/readings/arxiv-cs-ai-paper-says-review-and-deployment-constrained-6f5c0cc7.json"},"attribution":"Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.","maintainedGuideLinks":[],"readerContext":{"version":"reader-context-v1","attribution":"Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.","terms":[{"id":"term-llm","name":"Large language model","explanation":"A language model with many learned numerical settings, used to process and generate language.","referenceUrl":"https://developers.google.com/machine-learning/glossary#large-language-model"}],"entities":[{"id":"arxiv","name":"arXiv","kind":"organization","context":"arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.","referenceUrls":["https://arxiv.org/","https://info.arxiv.org/help/api/index.html"],"path":"/entities/arxiv","json":"/entities/arxiv.json","markdown":"/entities/arxiv.md"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"6f5c0cc70e5464280e79d201c9836a63e82516879f72ea26b6bacc61123af871","revision":"0074f38d-5a09-4682-858a-3f59511ab19e","revisionKind":"publication-id","title":"arXiv cs.AI paper says review and deployment constrained enterprise code migrations","url":"https://escapemeridian.com/readings/arxiv-cs-ai-paper-says-review-and-deployment-constrained-6f5c0cc7"},"source":{"type":"source","id":"https://arxiv.org/abs/2611.05457","revision":"7dfc14df7f48f71caa4383dba4132f7c2b3511204ef79f8940013c16c8bcf08e","revisionKind":"sha256-source-text","title":"Hybrid LLM-Deterministic Pipelines for Enterprise-Scale Code Migration at Bloomberg","url":"https://arxiv.org/abs/2611.05457"},"status":"available","items":[],"next":null,"scope":"Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter."},"revisions":[{"id":"6f5c0cc70e5464280e79d201c9836a63e82516879f72ea26b6bacc61123af871","title":"arXiv cs.AI paper says review and deployment constrained enterprise code migrations","argument":"Mario Corchero and coauthors report in an arXiv paper that Automated Uplift used LLM-assisted pipelines to migrate code across thousands of Bloomberg repositories. The multi-month initiative covered two campaigns: Python packaging and the C++ build system. Together, they generated 2,333 pull requests and resulted in 729 successful migrations.\n\nThe authors’ central finding is that producing proposed code changes was easier than getting them reviewed, tested, and deployed by the teams that own the repositories. Their hybrid architecture maximized deterministic processing, confined the LLM to bounded file transformations within a pre-computed context, and incorporated production telemetry such as execution logs and usage traces. The authors say costs for owning teams fell when changes could be verified in the pipeline before opening a pull request, as with C++ compilation and test validation, and when repositories were uniform enough for fixes to carry over.\n\nThe authors present these experiences as patterns for architecting hybrid pipelines and scaling code migrations. Their abstract reports outcomes from two campaigns; it does not establish how the approach would perform elsewhere.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The evidence is the arXiv abstract; it does not provide detailed campaign methods or results beyond the summary given there. The reported outcomes come from two Bloomberg campaigns and do not establish how the patterns would perform in other organizations or migrations.","speaker":{"name":"Corchero, Mario, Williams, David, Kollar, Laszlo Kiss, Liu, Haowen, Bairoliya, Shivam, Malbasic, Gala, Kirbas, Serkan, Sarro, Federica","roleAtSource":"Article author"},"source":{"title":"Hybrid LLM-Deterministic Pipelines for Enterprise-Scale Code Migration at Bloomberg","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2611.05457","publishedAt":"2026-11-03","dateBasis":"Date basis not provided."},"evidence":[{"text":"The paper reports two Automated Uplift campaigns, for Python packaging and a C++ build system, producing 2,333 pull requests and 729 successful migrations in total.","excerpt":"Together, these campaigns generated 2,333 pull requests, resulting in 729 successful migrations.","url":"https://arxiv.org/abs/2611.05457","relationship":"Source-stated"},{"text":"The authors identify review, testing, and deployment by repository-owning teams—not generating migrations—as the bottleneck.","excerpt":"generating migrations turned out to be the easier part; the bottleneck was the cost, for owning teams, of reviewing, testing, and deploying them.","url":"https://arxiv.org/abs/2611.05457","relationship":"Source-stated"},{"text":"The authors say costs fell when changes could be verified in the pipeline before a pull request and when fixes could carry over among sufficiently uniform repositories.","excerpt":"These costs were lowered when changes could be verified directly in the pipeline before opening a PR (as with C++ compilation and test validation), and when the population was uniform enough for fixes to carry over from one repository to the next.","url":"https://arxiv.org/abs/2611.05457","relationship":"Source-stated"},{"text":"The authors describe an architecture that favors deterministic processing and restricts the LLM to bounded file transformations using pre-computed context.","excerpt":"maximizes deterministic processing, confines the LLM to bounded file transformations within a pre-computed context","url":"https://arxiv.org/abs/2611.05457","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.05457","checkedAt":"2026-11-10T01:56:07.104Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How well would the authors’ patterns transfer to migrations beyond these two campaigns?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"arXiv cs.AI paper says review and deployment constrained enterprise code migrations","summary":"Mario Corchero and coauthors report in an arXiv paper that Automated Uplift used LLM-assisted pipelines to migrate code across thousands of Bloomberg repositories. The multi-month initiative covered two campaigns: Python packaging and the C++ build system. Together, they generated 2,333 pull requests and resulted in 729 successful migrations.\n\nThe authors’ central finding is that producing proposed code changes was easier than getting them reviewed, tested, and deployed by the teams that own the repositories. Their hybrid architecture maximized deterministic processing, confined the LLM to bounded file transformations within a pre-computed context, and incorporated production telemetry such as execution logs and usage traces. The authors say costs for owning teams fell when changes could be verified in the pipeline before opening a pull request, as with C++ compilation and test validation, and when repositories were uniform enough for fixes to carry over.\n\nThe authors present these experiences as patterns for architecting hybrid pipelines and scaling code migrations. Their abstract reports outcomes from two campaigns; it does not establish how the approach would perform elsewhere.","question":"How well would the authors’ patterns transfer to migrations beyond these two campaigns?","claims":[{"text":"The paper reports two Automated Uplift campaigns, for Python packaging and a C++ build system, producing 2,333 pull requests and 729 successful migrations in total.","excerpt":"Together, these campaigns generated 2,333 pull requests, resulting in 729 successful migrations.","support":"source-stated"},{"text":"The authors identify review, testing, and deployment by repository-owning teams—not generating migrations—as the bottleneck.","excerpt":"generating migrations turned out to be the easier part; the bottleneck was the cost, for owning teams, of reviewing, testing, and deploying them.","support":"source-stated"},{"text":"The authors say costs fell when changes could be verified in the pipeline before a pull request and when fixes could carry over among sufficiently uniform repositories.","excerpt":"These costs were lowered when changes could be verified directly in the pipeline before opening a PR (as with C++ compilation and test validation), and when the population was uniform enough for fixes to carry over from one repository to the next.","support":"source-stated"},{"text":"The authors describe an architecture that favors deterministic processing and restricts the LLM to bounded file transformations using pre-computed context.","excerpt":"maximizes deterministic processing, confines the LLM to bounded file transformations within a pre-computed context","support":"source-stated"}],"limitations":["The evidence is the arXiv abstract; it does not provide detailed campaign methods or results beyond the summary given there.","The reported outcomes come from two Bloomberg campaigns and do not establish how the patterns would perform in other organizations or migrations."],"mediaId":"8f4b0d0d78c3519749ff9a76c8dbbe9da0b66d1e4e1f2012890c6bca26adee0a","correctionReason":null,"id":"6f5c0cc70e5464280e79d201c9836a63e82516879f72ea26b6bacc61123af871","publicationId":"0074f38d-5a09-4682-858a-3f59511ab19e","version":1,"sourceUrl":"https://arxiv.org/abs/2611.05457","sourceTitle":"Hybrid LLM-Deterministic Pipelines for Enterprise-Scale Code Migration at Bloomberg","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-11-03","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-05T05:06:33.248Z","createdAt":"2026-11-10T01:25:10.085Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"8f4b0d0d78c3519749ff9a76c8dbbe9da0b66d1e4e1f2012890c6bca26adee0a","kind":"article","url":"https://arxiv.org/abs/2611.05457","sourceUrl":"https://arxiv.org/abs/2611.05457","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-10T01:24:18.362Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.05457","checkedAt":"2026-11-10T01:56:07.104Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"8f4b0d0d78c3519749ff9a76c8dbbe9da0b66d1e4e1f2012890c6bca26adee0a","confidence":0.81,"provider":"typesafe","model":"jev-1.13.0","inputHash":"57e21a7e8bac1bf11d557f55355e7aff3f271c3930615d094ced37ad62fc1411","decidedAt":"2026-11-10T01:24:20.353Z"},"provenance":{"textHash":"7dfc14df7f48f71caa4383dba4132f7c2b3511204ef79f8940013c16c8bcf08e","observationId":"a1996bdda4a9750ac5ae1d1884a8737b84309c472e9abb8c988e234a062dd478","retrievedAt":"2026-11-10T01:24:18.043Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"c6ee7c60-885d-4703-8d4b-ffd12c392631"},"reviewer":{"model":"gpt-6-luna","runId":"51a4fe9a-d35a-4834-8769-d6f08a9671a0"}},"audit":{"status":"passed","reviewedAt":"2026-11-10T01:25:10.085Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Corchero, Mario","url":null,"kind":"unknown"},{"name":"Williams, David","url":null,"kind":"unknown"},{"name":"Kollar, Laszlo Kiss","url":null,"kind":"unknown"},{"name":"Liu, Haowen","url":null,"kind":"unknown"},{"name":"Bairoliya, Shivam","url":null,"kind":"unknown"},{"name":"Malbasic, Gala","url":null,"kind":"unknown"},{"name":"Kirbas, Serkan","url":null,"kind":"unknown"},{"name":"Sarro, Federica","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Corchero, Mario","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Corchero, Mario","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05457","locator":"head/meta[name=author|citation_author]"},"assertionId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T01:56:07.104Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f","sourceUrl":"https://arxiv.org/abs/2611.05457","contentSha256":"d1c7572c9f145afb701333a91f30c32a8275ce7483243f6de546dd68ef9a0df4","textHash":"7dfc14df7f48f71caa4383dba4132f7c2b3511204ef79f8940013c16c8bcf08e","observedAt":"2026-11-10T01:56:07.104Z"}},{"actor":{"name":"Williams, David","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Williams, David","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05457","locator":"head/meta[name=author|citation_author]"},"assertionId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T01:56:07.104Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f","sourceUrl":"https://arxiv.org/abs/2611.05457","contentSha256":"d1c7572c9f145afb701333a91f30c32a8275ce7483243f6de546dd68ef9a0df4","textHash":"7dfc14df7f48f71caa4383dba4132f7c2b3511204ef79f8940013c16c8bcf08e","observedAt":"2026-11-10T01:56:07.104Z"}},{"actor":{"name":"Kollar, Laszlo Kiss","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Kollar, Laszlo Kiss","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05457","locator":"head/meta[name=author|citation_author]"},"assertionId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T01:56:07.104Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f","sourceUrl":"https://arxiv.org/abs/2611.05457","contentSha256":"d1c7572c9f145afb701333a91f30c32a8275ce7483243f6de546dd68ef9a0df4","textHash":"7dfc14df7f48f71caa4383dba4132f7c2b3511204ef79f8940013c16c8bcf08e","observedAt":"2026-11-10T01:56:07.104Z"}},{"actor":{"name":"Liu, Haowen","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Liu, Haowen","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05457","locator":"head/meta[name=author|citation_author]"},"assertionId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T01:56:07.104Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f","sourceUrl":"https://arxiv.org/abs/2611.05457","contentSha256":"d1c7572c9f145afb701333a91f30c32a8275ce7483243f6de546dd68ef9a0df4","textHash":"7dfc14df7f48f71caa4383dba4132f7c2b3511204ef79f8940013c16c8bcf08e","observedAt":"2026-11-10T01:56:07.104Z"}},{"actor":{"name":"Bairoliya, Shivam","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Bairoliya, Shivam","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05457","locator":"head/meta[name=author|citation_author]"},"assertionId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T01:56:07.104Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f","sourceUrl":"https://arxiv.org/abs/2611.05457","contentSha256":"d1c7572c9f145afb701333a91f30c32a8275ce7483243f6de546dd68ef9a0df4","textHash":"7dfc14df7f48f71caa4383dba4132f7c2b3511204ef79f8940013c16c8bcf08e","observedAt":"2026-11-10T01:56:07.104Z"}},{"actor":{"name":"Malbasic, Gala","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Malbasic, Gala","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05457","locator":"head/meta[name=author|citation_author]"},"assertionId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f:participation:5","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T01:56:07.104Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f","sourceUrl":"https://arxiv.org/abs/2611.05457","contentSha256":"d1c7572c9f145afb701333a91f30c32a8275ce7483243f6de546dd68ef9a0df4","textHash":"7dfc14df7f48f71caa4383dba4132f7c2b3511204ef79f8940013c16c8bcf08e","observedAt":"2026-11-10T01:56:07.104Z"}},{"actor":{"name":"Kirbas, Serkan","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Kirbas, Serkan","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05457","locator":"head/meta[name=author|citation_author]"},"assertionId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f:participation:6","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T01:56:07.104Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f","sourceUrl":"https://arxiv.org/abs/2611.05457","contentSha256":"d1c7572c9f145afb701333a91f30c32a8275ce7483243f6de546dd68ef9a0df4","textHash":"7dfc14df7f48f71caa4383dba4132f7c2b3511204ef79f8940013c16c8bcf08e","observedAt":"2026-11-10T01:56:07.104Z"}},{"actor":{"name":"Sarro, Federica","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Sarro, Federica","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.05457","locator":"head/meta[name=author|citation_author]"},"assertionId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f:participation:7","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T01:56:07.104Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f","sourceUrl":"https://arxiv.org/abs/2611.05457","contentSha256":"d1c7572c9f145afb701333a91f30c32a8275ce7483243f6de546dd68ef9a0df4","textHash":"7dfc14df7f48f71caa4383dba4132f7c2b3511204ef79f8940013c16c8bcf08e","observedAt":"2026-11-10T01:56:07.104Z"}}],"participationCoverage":{"total":8,"returned":8,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2611.05457","snapshotId":"b307f997c9fa9a7e3721a945c61d5b51dfcb1950991ba404611b596ae0155f7f","contentSha256":"d1c7572c9f145afb701333a91f30c32a8275ce7483243f6de546dd68ef9a0df4","textHash":"7dfc14df7f48f71caa4383dba4132f7c2b3511204ef79f8940013c16c8bcf08e","observedAt":"2026-11-10T01:56:07.104Z","basis":"publisher-page-metadata","sourceLinks":45,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/6f5c0cc70e5464280e79d201c9836a63e82516879f72ea26b6bacc61123af871/source"}}}],"audits":[{"publicationId":"0074f38d-5a09-4682-858a-3f59511ab19e","verdict":"passed","findings":[{"dimension":"attribution","status":"pass","explanation":"The reading attributes the findings to Mario Corchero and coauthors, matching the listed authors. It keeps the bottleneck and cost findings as the authors' statements. Calling it an 'arXiv cs.AI paper' is acceptable: the primary category is cs.SE, but the paper is cross-listed in cs.AI, which is the supplied publisher.","excerpt":"Subjects:\nSoftware Engineering (cs.SE); Artificial Intelligence (cs.AI)"},{"dimension":"summary","status":"pass","explanation":"The reading faithfully conveys the actor, the two campaigns, the counts, the hybrid design and the central argument that review and deployment were the bottleneck. Two small repairs are optional. First, 'migrate code across thousands of repositories' could be read as thousands of completed migrations; 'applied across thousands of candidate repositories' fits the source better. Second, noting that the Python campaign was heterogeneous and the C++ campaign uniform would sharpen the uniformity point. Neither prevents understanding.","excerpt":"a heterogeneous Python packaging migration (approx. 1,900 candidate repositories) and a uniform C++ build system migration (approx. 6,700 candidate repositories)"},{"dimension":"date","status":"pass","explanation":"The original publication date of 2026-11-03 matches the v1 submission date in the source, not the later feed date.","excerpt":"[Submitted on 3 Nov 2026]"},{"dimension":"support","status":"pass","explanation":"All four claims are directly stated in the abstract. The limitation that the results come from two Bloomberg campaigns is properly conditional, and the transfer question follows from the evidence. The authors' own description of the patterns as 'actionable, transferable' is softened rather than misrepresented; this is not a fidelity failure.","excerpt":"actionable, transferable patterns for architecting hybrid LLM-deterministic pipelines and scaling code migrations across the enterprise"},{"dimension":"media","status":"pass","explanation":"The selected media is the verified direct link to the original arXiv abstract page, which is the most relevant source.","excerpt":null}],"createdAt":"2026-11-10T01:25:10.085Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}