{"id":51475362,"url":"https://github.com/asmuelle/selene","last_synced_at":"2026-07-06T20:30:33.274Z","repository":{"id":366157023,"uuid":"1266479263","full_name":"asmuelle/selene","owner":"asmuelle","description":"A cycle, fertility, and perimenopause tracker whose AI insight engine runs entirely on-device — no server to subpoena, breach, or sell data from, verifiable in airplane mode.","archived":false,"fork":false,"pushed_at":"2026-06-20T13:23:22.000Z","size":207,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-06-20T15:17:01.289Z","etag":null,"topics":["android","cycle-tracking","edge-ai","femtech","fertility","ios","on-device-inference","privacy"],"latest_commit_sha":null,"homepage":"https://asmuelle.github.io/selene/","language":"Swift","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/asmuelle.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":"SECURITY.md","support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":"AGENTS.md","dco":null,"cla":null}},"created_at":"2026-06-11T16:50:28.000Z","updated_at":"2026-06-20T13:23:25.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/asmuelle/selene","commit_stats":null,"previous_names":["asmuelle/selene"],"tags_count":null,"template":false,"template_full_name":null,"purl":"pkg:github/asmuelle/selene","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/asmuelle%2Fselene","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/asmuelle%2Fselene/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/asmuelle%2Fselene/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/asmuelle%2Fselene/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/asmuelle","download_url":"https://codeload.github.com/asmuelle/selene/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/asmuelle%2Fselene/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":35205739,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-26T15:22:16.424Z","status":"online","status_checked_at":"2026-07-06T02:00:07.184Z","response_time":106,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["android","cycle-tracking","edge-ai","femtech","fertility","ios","on-device-inference","privacy"],"created_at":"2026-07-06T20:30:32.518Z","updated_at":"2026-07-06T20:30:33.267Z","avatar_url":"https://github.com/asmuelle.png","language":"Swift","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Selene\n\n[![CI](https://github.com/asmuelle/selene/actions/workflows/ci.yml/badge.svg)](https://github.com/asmuelle/selene/actions/workflows/ci.yml)\n\n\u003e A cycle, fertility, and perimenopause tracker whose AI insight engine runs entirely on-device — no server to subpoena, breach, or sell data from, verifiable in airplane mode.\n\n## Concept\n\nA cycle, fertility, and perimenopause tracker whose AI insight engine runs entirely on-device — no server to subpoena, breach, or sell data from, verifiable in airplane mode.\n\n## Target User\n\nWomen 18-45 who churned from Flo/Clue after the $59.5M settlements and post-Dobbs subpoena fears; secondary wedge: perimenopausal women 35-55 ignored by fertility-focused incumbents.\n\n## Why Edge AI Is Structural (not decoration)\n\niOS: AFM 3 Core (3B) with @Generable guided generation turns free-text/voice symptom logs into structured entries; Dynamic Profiles swap between logging, cycle-forecast, and fertility-Q\u0026A modes; AFM 3 Core Advanced image input (iPhone 15 Pro+) reads ovulation/pregnancy test strips from photos. Android: Gemini Nano Prompt API on flagships, bundled Gemma 3n E2B via LiteRT-LM on mid-range. Structural, not decorative: reproductive data in any cloud is now a legal liability (Meta CIPA verdict, discovery requests), so the core promise — 'we cannot hand over what we never have' — is only true with zero data egress, proven via airplane-mode operation and published network audits.\n\n## Why Now (2026 timing)\n\nTrust in cloud period trackers collapsed in 2025-26 while the category remains a proven mass market; WWDC26's AFM 3 image input just made on-device test-strip reading possible; Guideline 5.1.2(i) consent friction now actively penalizes cloud-AI competitors at App Review.\n\n## Tech Stack\n\niOS (iOS 26+, iPhone 15 Pro+ floor): SwiftUI; FoundationModels framework — SystemLanguageModel with @Generable guided generation for free-text/voice symptom → structured entry, Tool calling into a local stats engine; deterministic hierarchical Bayesian cycle/ovulation model in Swift (no LLM) for forecasts; SpeechAnalyzer/SpeechTranscriber (iOS 26) for on-device ASR instead of Whisper; test strips via a dedicated compact Core ML vision model (Vision framework strip detection + custom CNN classifier trained on LH/hCG strip datasets) — do NOT depend on AFM 3 Core Advanced image input given its high-end-only device floor; local RAG over a curated, citation-pinned ACOG/NICE-derived content pack using a small on-device embedding model (EmbeddingGemma-class via Core ML) to ground fertility/perimenopause Q\u0026A and cap hallucination; GRDB/SQLite with NSFileProtectionComplete, backup exclusion by default + Advanced Data Protection guidance; StoreKit 2 only network surface; zero third-party SDKs; publish reproducible network audits (mitmproxy capture + App Privacy Report). Android: Kotlin + Jetpack Compose; ML Kit GenAI Prompt API (Gemini Nano) on supported flagships; fallback Gemma 3n E2B / Gemma 4-class via MediaPipe LLM Inference API on LiteRT-LM, weights delivered through Play Feature Delivery (on-demand, not in base APK); whisper.cpp small or offline SpeechRecognizer for ASR; same strip-reading model exported to LiteRT; Room + SQLCipher. Shared: Kotlin Multiplatform core for cycle math, Bayesian forecaster, schema, and content pack so insight logic is identical across platforms; ship Android as logging-first with AI tier gated to Nano-capable devices at launch.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fasmuelle%2Fselene","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fasmuelle%2Fselene","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fasmuelle%2Fselene/lists"}