[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$fRXOtHeuIyJNo3nL6H-fk4F9JDuhsfGk9qjJ3qMoEfvA":3,"$fC5vW9iaJJA4RWDeshQUHwFXl8h1c9lt_IYWzQVaVpWI":6},{"success":4,"data":5},true,{},{"data":7},[8,40,71,96,122,145,159,173,185,196,208,220,232,243,253,273,284,294,303,316,336,350,361,370,379,393],{"id":9,"title":10,"description":11,"long_description":12,"image":13,"image_url":14,"technologies":15,"category":22,"github_url":23,"demo_url":24,"featured":4,"status":25,"created_at":26,"updated_at":26,"problem":27,"approach":28,"trouble":29,"result_metric":30,"result_label":31,"role":32,"year":13,"tier":33,"link_status":34,"title_ko":13,"description_ko":35,"problem_ko":36,"approach_ko":37,"trouble_ko":38,"result_label_ko":39},"8b19a309-86d8-493d-9a5a-8e46bbdda696","scenefile","My first npm package: an open-source CLI that turns a short JSON file into an MP4, built so any language model, even one that cannot see images, can make a correct video and fix its own mistakes.","Twelve scene types, ten themes, bundled fonts for Latin and Cyrillic, narration in English and Russian generated on your own machine, captions and a music bed. Its check command lays every scene out in a real browser and reports each problem as text with its fix. An MCP server, recipe skills and a Claude Code plugin let coding agents use it directly.",null,"https:\u002F\u002Fkalandarov.uz\u002Fmedia\u002Fwork\u002Fscenefile.jpg",[16,17,18,19,20,21],"TypeScript","Node.js","Headless browser","ffmpeg","MCP","Vitest","AI","https:\u002F\u002Fgithub.com\u002FXusniddin-devv\u002Fscenefile","https:\u002F\u002Fwww.npmjs.com\u002Fpackage\u002Fscenefile","Live","2026-10-14T08:25:03.956596+00:00","Language models can write code but cannot see a video, so they cannot tell when text overflows a frame or a scene runs too long. Video tools built for people give them nothing to read back.","A scenefile is JSON with typed scenes. The engine measures each scene in a real browser and its check command turns every problem into a message with the field's path and a fix, so a model can correct itself in a loop. Rendering runs in several browsers at once and ffmpeg assembles the result; the guide models read is generated from the scene types, and a test fails if it drifts.","On Linux with the ffmpeg Ubuntu ships, a clip after the first scene played at the wrong moment, and it never showed on macOS where the tool was built. Footage is now placed in ffmpeg's own time units whatever its version, and CI renders clips on Linux.","11\u002F12","scenefiles from four inexpensive models passed the full check on the first reply (informal trial, Oct 2026)","Solo build","case","live","첫 npm 패키지. 짧은 JSON 파일을 MP4 영상으로 바꾸는 오픈소스 CLI로, 이미지를 볼 수 없는 언어 모델도 올바른 영상을 만들고 스스로 오류를 고칠 수 있도록 설계했습니다.","언어 모델은 코드를 쓸 수 있지만 영상을 볼 수 없어서, 글자가 화면을 벗어나거나 장면이 너무 길어도 알아차리지 못합니다. 사람을 위한 영상 도구는 모델이 읽을 수 있는 피드백을 주지 않습니다.","scenefile은 장면 타입이 정해진 JSON입니다. 엔진이 실제 브라우저에서 각 장면의 레이아웃을 측정하고, check 명령은 모든 문제를 필드 경로와 해결 방법이 담긴 메시지로 돌려주어 모델이 반복하며 스스로 고칠 수 있습니다. 렌더링은 여러 브라우저에서 병렬로 실행되고 ffmpeg가 영상을 합칩니다. 모델이 읽는 가이드는 장면 타입에서 자동 생성되며, 둘이 어긋나면 테스트가 실패합니다.","Ubuntu에 기본 포함된 ffmpeg를 쓰는 Linux에서는 첫 장면 이후의 클립이 엉뚱한 시점에 재생되었고, 개발 환경인 macOS에서는 이 문제가 드러나지 않았습니다. 이제 영상 배치는 버전과 관계없이 ffmpeg 자체의 시간 단위로 계산되며, CI가 Linux에서 클립을 직접 렌더링합니다.","저가 모델 네 개가 작성한 scenefile 중 첫 응답에서 전체 검사를 통과한 비율 (비공식 시험, 2026년 10월)",{"id":41,"title":42,"description":43,"long_description":44,"image":13,"image_url":45,"technologies":46,"category":54,"github_url":13,"demo_url":55,"featured":4,"status":56,"created_at":57,"updated_at":58,"problem":59,"approach":60,"trouble":61,"result_metric":62,"result_label":63,"role":64,"year":13,"tier":33,"link_status":34,"title_ko":65,"description_ko":66,"problem_ko":67,"approach_ko":68,"trouble_ko":69,"result_label_ko":70},"6691fc24-8305-4b2f-ba85-ccf6c9ef6c3b","MoykaGo","Car-wash marketplace for Uzbekistan: drivers find a free bay nearby and book it, owners run the queue.","Two-sided marketplace. Drivers search for a wash bay by distance and availability; owners manage capacity, pricing and a live queue. Nearest-wash matching runs on PostGIS geography(Point,4326); the queue pushes state to both apps over SSE; billing is subscription-first with three plan states rather than per-minute charging.","https:\u002F\u002Fkalandarov.uz\u002Fmedia\u002Fwork\u002Fmoykago.jpg",[47,48,49,50,51,52,53],"Fastify","PostgreSQL","PostGIS","Kysely","Expo","React Native","SSE","Full Stack","https:\u002F\u002Fmoyka-go.uz","Completed","2026-09-10T08:30:04.940981+00:00","2026-09-21T07:50:29.722+00:00","Drivers had no way to find a free car-wash bay nearby, and owners ran their queue on paper and phone calls.","Two-sided marketplace: nearest-wash matching on PostGIS geography(Point,4326), a live queue pushed to both apps over SSE, and subscription-first billing with three plan states instead of per-minute charging.","Per-minute pricing was the original model and it made the exhausted-plan state unrepresentable — a driver could start a wash the system could not bill. Moving to three explicit subscription states made the failure case a state rather than an edge case.","1,205","server test cases across 87 files","Solo build — backend, mobile, infrastructure","","우즈베키스탄 세차 마켓플레이스. 운전자는 근처의 빈 세차 베이를 찾아 예약하고, 사업주는 대기열을 관리합니다.","운전자에게는 근처에 비어 있는 세차 베이를 찾을 방법이 없었고, 사업주는 대기열을 종이와 전화로 관리하고 있었습니다.","양면 마켓플레이스로 구현했습니다. 최근접 세차장 매칭은 PostGIS geography(Point,4326) 위에서 동작하고, 대기열 상태는 SSE로 양쪽 앱에 실시간 전달되며, 과금은 분 단위 대신 세 가지 플랜 상태를 가진 구독 방식입니다.","처음 모델은 분 단위 과금이었는데, 이 구조에서는 '플랜 소진' 상태를 표현할 수 없었습니다. 결제가 불가능한 상태인데도 운전자가 세차를 시작할 수 있었던 것입니다. 구독 상태를 세 가지로 명시하면서, 이 실패 케이스가 예외가 아니라 하나의 상태로 다뤄지게 되었습니다.","87개 파일에 걸친 서버 테스트 케이스",{"id":72,"title":73,"description":74,"long_description":75,"image":13,"image_url":13,"technologies":76,"category":82,"github_url":13,"demo_url":83,"featured":4,"status":25,"created_at":84,"updated_at":85,"problem":86,"approach":87,"trouble":88,"result_metric":89,"result_label":90,"role":32,"year":13,"tier":33,"link_status":34,"title_ko":13,"description_ko":91,"problem_ko":92,"approach_ko":93,"trouble_ko":94,"result_label_ko":95},"40ad4f9c-6bec-4032-a9ad-71eb31d7ea9d","JobHunterKR","Job search for anyone looking for work in Korea: postings from 11 sources that allow collection, in one English feed with one set of filters.","Search and filters across ~6,500 open roles, company pages, an application tracker, and saved searches that alert by email or Telegram. CVs are read in the browser and never uploaded.",[77,78,79,80,81,21],"Next.js","Supabase (Postgres, Auth)","pg_cron","Netlify Functions","Resend","Web App","https:\u002F\u002Fjobhunterkr.kalandarov.uz","2026-07-20T09:00:00+00:00","2026-09-14T01:00:00+00:00","Korea's postings are spread across dozens of boards and career pages, mostly Korean-only, and the biggest boards forbid collection. 88% of employers publish only a Korean name, so searching \"kakao\" or \"samsung\" in English finds almost nothing, and every board uses its own categories.","11 permitted sources (Linkareer, Jumpit, KOWORK, K-Campus, GreetingHR, Ninehire, Greenhouse\u002FLever\u002FAshby, Kakao, NAVER, LINE, 알바몬) are collected hourly into Postgres; the rest are deep links. One taxonomy for all of them (20 fields, 17 provinces, level, years, contract, language); every Korean title shown in English beside the original; employer aliases let English search reach Korean-only names. Saved searches alert within minutes by email or Telegram, driven by pg_cron.","Supabase returns at most 1,000 rows per request, with no error. Ingest read stored listings in one request and saw 1,000 of 5,808, so postings taken down outside that window never counted as missing: 1,229 dead links stayed open, and the alert matcher skipped 76% of new jobs the same way. Paging every read fixed it; the first run after closed 57 stale postings.","6567","open roles from 1,678 employers, every title readable in English (Sept 2026)","한국에서 일자리를 찾는 모든 사람을 위한 채용 검색 서비스. 수집이 허용된 11개 소스의 공고를 하나의 영어 피드와 하나의 필터로 모았습니다.","한국의 채용 공고는 수십 개의 채용 사이트와 기업 채용 페이지에 흩어져 있고 대부분 한국어로만 되어 있으며, 규모가 큰 사이트는 수집을 금지합니다. 기업의 88%가 한국어 이름만 공개하기 때문에 영어로 \"kakao\"나 \"samsung\"을 검색하면 거의 아무것도 나오지 않고, 사이트마다 직무 분류도 제각각입니다.","수집이 허용된 11개 소스(Linkareer, Jumpit, KOWORK, K-Campus, GreetingHR, Ninehire, Greenhouse\u002FLever\u002FAshby, 카카오, 네이버, LINE, 알바몬)를 매시간 Postgres에 모으고, 나머지는 딥링크로 연결합니다. 모든 소스를 하나의 분류 체계(직무 20개, 17개 시도, 경력 수준, 요구 경력 연수, 고용 형태, 언어)로 정리하고, 한국어 제목은 원문 옆에 영어로 보여 주며, 기업 별칭 덕분에 영어 검색으로도 한국어 이름만 있는 기업을 찾을 수 있습니다. 저장한 검색 조건은 pg_cron으로 몇 분 안에 이메일이나 텔레그램으로 알림을 보냅니다.","Supabase는 한 번의 요청에 최대 1,000행만 반환하며 오류도 내지 않습니다. 수집 작업이 저장된 공고를 한 번의 요청으로 읽어 5,808건 중 1,000건만 보았기 때문에, 그 범위 밖에서 내려간 공고는 사라진 것으로 집계되지 않았습니다. 그 결과 죽은 링크 1,229건이 열린 공고로 남았고, 알림 매칭도 같은 이유로 새 공고의 76%를 건너뛰었습니다. 모든 읽기를 페이지 단위로 나누어 해결했고, 수정 후 첫 실행에서 오래된 공고 57건이 마감 처리되었습니다.","2026년 9월 기준, 1,678개 기업의 채용 중인 공고, 모든 제목을 영어로 확인 가능",{"id":97,"title":98,"description":99,"long_description":100,"image":101,"image_url":102,"technologies":103,"category":22,"github_url":13,"demo_url":109,"featured":4,"status":25,"created_at":110,"updated_at":111,"problem":112,"approach":113,"trouble":114,"result_metric":115,"result_label":116,"role":32,"year":13,"tier":33,"link_status":34,"title_ko":13,"description_ko":117,"problem_ko":118,"approach_ko":119,"trouble_ko":120,"result_label_ko":121},"fbb6414f-654e-4556-99aa-f5a4ddb4c62b","Hannim","Korean + TOPIK II coaching web app with reader, exam, tutor, listening, voice and OCR.","Two-profile Korean learning platform: graded reader, TOPIK II mock exams, AI tutor, listening drills, conjugation trainer, mistakes bank and tap-to-define. RAG-grounded feedback over pgvector.","https:\u002F\u002Fatpyngridfrvfrbwkyyu.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fproject-images\u002Fhannim.jpeg","https:\u002F\u002Fkalandarov.uz\u002Fmedia\u002Fwork\u002Fhannim.jpg",[104,105,106,107,108,48],"Nuxt","pgvector","OpenRouter","Groq","Cohere","https:\u002F\u002Fwww.hannim.app","2026-07-19T10:55:07.805217+00:00","2026-07-19T15:56:31.896+00:00","TOPIK II learners get either a graded reader or an AI tutor, never both grounded in the same corpus, so feedback contradicts the material.","Forty API routes over a single pgvector store — reader, mock exams, tutor, listening drills, conjugation practice and a mistakes bank all retrieve against the same embedded corpus.","Embedding the corpus hit Cohere's 100k tokens\u002Fminute throttle mid-ingest and failed partway, leaving the store inconsistent. Batching the work off-peak and making ingest resumable fixed both the throttle and the partial-state problem.","40","LLM routes over one pgvector corpus","TOPIK II 학습자를 위한 한국어 학습 플랫폼. 읽기 자료, 모의고사, AI 튜터, 듣기 훈련을 하나의 코퍼스 위에서 제공합니다.","TOPIK II 학습자는 난이도별 읽기 자료 아니면 AI 튜터 중 하나만 쓰게 되고, 둘이 같은 코퍼스에 기반하지 않다 보니 피드백이 학습 자료와 어긋납니다.","하나의 pgvector 저장소 위에 40개의 API 라우트를 올렸습니다. 읽기, 모의고사, 튜터, 듣기, 활용 연습, 오답 노트가 모두 동일한 임베딩 코퍼스를 참조합니다.","코퍼스 임베딩 도중 Cohere의 분당 10만 토큰 제한에 걸려 작업이 중간에 실패했고, 저장소가 불완전한 상태로 남았습니다. 작업을 비피크 시간대로 배치하고 재개 가능하도록 만들면서 처리량 제한과 부분 상태 문제를 함께 해결했습니다.","하나의 pgvector 코퍼스 위의 LLM 라우트",{"id":123,"title":124,"description":125,"long_description":126,"image":127,"image_url":127,"technologies":128,"category":22,"github_url":13,"demo_url":132,"featured":4,"status":25,"created_at":133,"updated_at":134,"problem":135,"approach":136,"trouble":137,"result_metric":138,"result_label":139,"role":32,"year":13,"tier":33,"link_status":34,"title_ko":65,"description_ko":140,"problem_ko":141,"approach_ko":142,"trouble_ko":143,"result_label_ko":144},"2fa53735-1c8d-4b71-89bd-62aa265756c1","Law4Kor","RAG legal-information system that answers foreign students' questions about Korean law with citations.","Retrieval-augmented legal assistant for foreign students in Korea. Grounds every answer in real statutes via pgvector, streams responses over SSE, deployed behind Caddy.","https:\u002F\u002Fatpyngridfrvfrbwkyyu.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fproject-images\u002Fkorean-law-rag.jpeg",[129,105,130,106,131],"RAG","FastAPI","Caddy","https:\u002F\u002Flaw4kor.uz","2026-07-18T10:55:07.805217+00:00","2026-09-21T07:50:15.855+00:00","Foreign students in Korea face legal questions where a wrong answer is expensive, and machine-translated statute is not trustworthy enough to act on.","Hybrid retrieval over a Korean statute corpus: dense pgvector search and sparse search fused with reciprocal rank fusion, then a calibrated reranker. Answers in plain English, every claim linked back to its source article.","The scope filter silently let the wrong statute through. Authority-lane hits arrive in two shapes — a statute ARTICLE carries statute_name_ko, but a statute-tier CHUNK (an English KLRI translation) has statute_name_ko=None and its name in title. Matching on the Korean name alone meant a Commercial Building Lease Protection Act chunk survived a residential-scope query untouched. Reproduced it before fixing, then matched on both shapes.","332","tests across the retrieval and answer pipeline","한국 거주 외국인 학생을 위한 법률 정보 RAG 시스템. 한국 법령을 근거와 함께 쉬운 영어로 설명합니다.","한국에 있는 외국인 학생은 답을 잘못 알면 대가가 큰 법률 문제를 마주하는데, 기계 번역된 법령은 그대로 믿고 행동하기에는 신뢰하기 어렵습니다.","한국 법령 코퍼스 위의 하이브리드 검색입니다. pgvector 기반 dense 검색과 sparse 검색을 RRF로 융합한 뒤 캘리브레이션된 reranker를 거칩니다. 답변은 쉬운 영어로 제공하고, 모든 주장은 원 조문으로 링크됩니다.","범위 필터가 엉뚱한 법령을 조용히 통과시키고 있었습니다. authority 레인의 검색 결과는 두 가지 형태로 들어오는데, 법령 ARTICLE은 statute_name_ko를 갖지만 statute-tier CHUNK(영문 KLRI 번역본)는 statute_name_ko가 None이고 이름이 title에 있습니다. 한국어 이름만으로 매칭하다 보니 상가건물 임대차보호법 청크가 주거 범위 질의에서 그대로 살아남았습니다. 먼저 재현한 뒤 두 형태 모두에 대해 매칭하도록 고쳤습니다.","검색·답변 파이프라인 전반의 테스트",{"id":146,"title":147,"description":148,"long_description":149,"image":150,"image_url":150,"technologies":151,"category":82,"github_url":13,"demo_url":155,"featured":4,"status":25,"created_at":156,"updated_at":156,"problem":13,"approach":13,"trouble":13,"result_metric":13,"result_label":13,"role":13,"year":13,"tier":157,"link_status":34,"title_ko":13,"description_ko":158,"problem_ko":13,"approach_ko":13,"trouble_ko":13,"result_label_ko":13},"dff8ded9-1741-4bb1-8bfa-e94d3f1f57da","TraWell","Tashkent tour-agency website with an inquiry-driven conversion flow and cinematic motion.","Rebranded travel-agency site with a photo-led product UI, GSAP motion design and a Supabase-backed inquiry funnel. Deployed on Netlify.","https:\u002F\u002Fatpyngridfrvfrbwkyyu.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fproject-images\u002Ftrawell.jpeg",[104,152,153,154],"Supabase","GSAP","Netlify","https:\u002F\u002Ftrawell-uz.netlify.app","2026-07-17T10:55:07.805217+00:00","index","타슈켄트 여행사 웹사이트. 문의 중심 전환 플로우와 시네마틱 모션을 적용했습니다.",{"id":160,"title":161,"description":162,"long_description":163,"image":13,"image_url":164,"technologies":165,"category":54,"github_url":13,"demo_url":169,"featured":4,"status":56,"created_at":170,"updated_at":171,"problem":13,"approach":13,"trouble":13,"result_metric":13,"result_label":13,"role":13,"year":13,"tier":157,"link_status":34,"title_ko":13,"description_ko":172,"problem_ko":13,"approach_ko":13,"trouble_ko":13,"result_label_ko":13},"168622ca-4926-4412-87fa-7da1529d385d","GroupWrapped","Viral Telegram group-roast bot and Mini App that analyzes chat activity.","Telegram bot (@groupanalyzebot) plus Mini App that roasts and 'wraps' group chats. Postgres-backed, deployed on Heroku, with a built-in payments layer.","https:\u002F\u002Fatpyngridfrvfrbwkyyu.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fblog-images\u002Fproject-1784460926691-qag09o14q0n.png",[17,166,48,167,168],"Telegram Bot API","DeepSeek","Heroku","https:\u002F\u002Ft.me\u002Fgroupanalyzebot","2026-07-16T10:55:07.805217+00:00","2026-07-19T11:35:53.048+00:00","텔레그램 단체 대화를 분석해 재미있게 정리해 주는 봇과 Mini App.",{"id":174,"title":175,"description":176,"long_description":177,"image":178,"image_url":178,"technologies":179,"category":82,"github_url":13,"demo_url":181,"featured":4,"status":25,"created_at":182,"updated_at":183,"problem":13,"approach":13,"trouble":13,"result_metric":13,"result_label":13,"role":13,"year":13,"tier":157,"link_status":34,"title_ko":13,"description_ko":184,"problem_ko":13,"approach_ko":13,"trouble_ko":13,"result_label_ko":13},"59fe2ca6-0bf1-423d-8975-21b5f6499938","FootHost","Football-pitch booking platform for Tashkent — reserve a field and pay a deposit online.","Find a stadium near you, assemble a team and pay a deposit via Click, Payme or Rahmat — no calls, no waiting. Includes tournaments, match lobbies and player ratings.","https:\u002F\u002Fatpyngridfrvfrbwkyyu.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fproject-images\u002Ffoothost.jpeg",[77,16,180],"Tailwind CSS","https:\u002F\u002Ffoothost-web.netlify.app","2026-07-16T08:17:25.770148+00:00","2026-07-19T11:17:25.770148+00:00","타슈켄트 축구장 예약 플랫폼. 구장을 예약하고 예약금을 온라인으로 결제합니다.",{"id":186,"title":187,"description":188,"long_description":189,"image":190,"image_url":190,"technologies":191,"category":82,"github_url":13,"demo_url":193,"featured":4,"status":25,"created_at":194,"updated_at":183,"problem":13,"approach":13,"trouble":13,"result_metric":13,"result_label":13,"role":13,"year":13,"tier":157,"link_status":34,"title_ko":13,"description_ko":195,"problem_ko":13,"approach_ko":13,"trouble_ko":13,"result_label_ko":13},"d6caa352-316e-41ce-990b-534a8a105d35","Sening Toy","Wedding e-invitation builder for Uzbek toy celebrations, with a drag-and-drop canvas editor.","Create and share custom digital wedding invitations: a Fabric.js canvas editor with GSAP motion, tailored for Uzbek weddings.","https:\u002F\u002Fatpyngridfrvfrbwkyyu.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fproject-images\u002Feinvite.jpeg",[77,192,153,16],"Fabric.js","https:\u002F\u002Fe-invite-uz.netlify.app","2026-07-16T06:17:25.770148+00:00","우즈베키스탄 결혼식(toy) 전자 청첩장 빌더. 드래그 앤드 드롭 캔버스 편집기를 제공합니다.",{"id":197,"title":198,"description":199,"long_description":200,"image":201,"image_url":201,"technologies":202,"category":82,"github_url":13,"demo_url":204,"featured":205,"status":25,"created_at":206,"updated_at":183,"problem":13,"approach":13,"trouble":13,"result_metric":13,"result_label":13,"role":13,"year":13,"tier":157,"link_status":34,"title_ko":13,"description_ko":207,"problem_ko":13,"approach_ko":13,"trouble_ko":13,"result_label_ko":13},"565715b7-ea0a-4149-a15b-cc9a552ad492","Stadium Map","Interactive map of 165 football stadiums (184 pitches) across Tashkent, by district.","Browse every football pitch in Tashkent on a Yandex-powered map, filtered by district — 184 fields across 165 stadiums.","https:\u002F\u002Fatpyngridfrvfrbwkyyu.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fproject-images\u002Fstadium-map.jpeg",[77,16,203],"Yandex Maps","https:\u002F\u002Fstadiumlist.netlify.app",false,"2026-07-16T04:17:25.770148+00:00","타슈켄트 전역 축구장 165곳(구장 184면)을 구별로 볼 수 있는 인터랙티브 지도.",{"id":209,"title":210,"description":211,"long_description":212,"image":213,"image_url":213,"technologies":214,"category":82,"github_url":13,"demo_url":216,"featured":205,"status":25,"created_at":217,"updated_at":218,"problem":13,"approach":13,"trouble":13,"result_metric":13,"result_label":13,"role":13,"year":13,"tier":157,"link_status":34,"title_ko":13,"description_ko":219,"problem_ko":13,"approach_ko":13,"trouble_ko":13,"result_label_ko":13},"1a5dbc14-96bb-4c29-a1c6-69e5cc2d0a49","UniBridge","Study-abroad consultancy helping Uzbek students into top Korean universities.","Consulting-agency site for Uzbek students applying to top Korean universities — service packages with a money-back guarantee, success stories and free-consultation booking, in Uzbek and English.","https:\u002F\u002Fatpyngridfrvfrbwkyyu.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fproject-images\u002Funibridge.jpeg",[77,215,180],"i18n","https:\u002F\u002Fconsulting-uzz.netlify.app","2026-07-16T02:23:30.769478+00:00","2026-07-19T11:23:30.769478+00:00","유학 컨설팅 서비스. 우즈베키스탄 학생의 한국 주요 대학 진학을 돕습니다.",{"id":221,"title":222,"description":223,"long_description":224,"image":225,"image_url":225,"technologies":226,"category":22,"github_url":13,"demo_url":229,"featured":205,"status":25,"created_at":230,"updated_at":218,"problem":13,"approach":13,"trouble":13,"result_metric":13,"result_label":13,"role":13,"year":13,"tier":157,"link_status":34,"title_ko":13,"description_ko":231,"problem_ko":13,"approach_ko":13,"trouble_ko":13,"result_label_ko":13},"a39fab82-25d8-49f2-9c7d-103e5052f7d0","Toto Planner","AI task planner that turns voice or text into scheduled tasks, with daily stats.","Personal AI planner: capture tasks by voice or text, see pending\u002Fdone stats, and get a smart daily plan. Mobile-first PWA.","https:\u002F\u002Fatpyngridfrvfrbwkyyu.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fproject-images\u002Ftoto-planner.jpeg",[227,22,228],"React","PWA","https:\u002F\u002Ftoto-planner-ai-v1.netlify.app","2026-07-16T00:23:30.769478+00:00","음성이나 텍스트를 일정으로 바꿔 주는 AI 태스크 플래너. 일간 통계도 제공합니다.",{"id":233,"title":234,"description":235,"long_description":236,"image":237,"image_url":237,"technologies":238,"category":82,"github_url":13,"demo_url":239,"featured":205,"status":25,"created_at":240,"updated_at":241,"problem":13,"approach":13,"trouble":13,"result_metric":13,"result_label":13,"role":13,"year":13,"tier":157,"link_status":34,"title_ko":13,"description_ko":242,"problem_ko":13,"approach_ko":13,"trouble_ko":13,"result_label_ko":13},"4bfb070f-4c94-4969-83d0-7ed18b759c64","Mekan Games","Gaming platform for schools where students sign in to play a curated game library.","A school-focused gaming platform: students log into a dashboard and play from a curated game library, with authentication and a games catalog.","https:\u002F\u002Fatpyngridfrvfrbwkyyu.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fproject-images\u002Fmekan.jpeg",[77,227,180],"https:\u002F\u002Fgame-dev-mekan.netlify.app","2026-07-15T22:43:35.782538+00:00","2026-07-19T11:43:35.782538+00:00","학교용 게임 플랫폼. 학생이 로그인해 선별된 게임 라이브러리를 이용합니다.",{"id":244,"title":245,"description":246,"long_description":247,"image":248,"image_url":248,"technologies":249,"category":82,"github_url":13,"demo_url":250,"featured":205,"status":25,"created_at":251,"updated_at":241,"problem":13,"approach":13,"trouble":13,"result_metric":13,"result_label":13,"role":13,"year":13,"tier":157,"link_status":34,"title_ko":13,"description_ko":252,"problem_ko":13,"approach_ko":13,"trouble_ko":13,"result_label_ko":13},"29385f7d-a891-4c91-aba6-215f2b4cc067","ManageBirthday","Birthday reminder app that tracks important dates and sends timely reminders.","Keep track of friends' and family's birthdays and get reminded in time, with a companion Telegram bot for notifications.","https:\u002F\u002Fatpyngridfrvfrbwkyyu.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fproject-images\u002Fmanagebirthday.jpeg",[227,16,166],"https:\u002F\u002Fbirthday-manager-app.netlify.app","2026-07-15T20:43:35.782538+00:00","생일 리마인더 앱. 중요한 날짜를 관리하고 제때 알림을 보냅니다.",{"id":254,"title":255,"description":256,"long_description":257,"image":13,"image_url":13,"technologies":258,"category":22,"github_url":13,"demo_url":13,"featured":4,"status":56,"created_at":261,"updated_at":261,"problem":262,"approach":263,"trouble":264,"result_metric":265,"result_label":266,"role":32,"year":13,"tier":33,"link_status":267,"title_ko":13,"description_ko":268,"problem_ko":269,"approach_ko":270,"trouble_ko":271,"result_label_ko":272},"9a21e360-fe96-448e-8c3c-649acfb91d4b","Boardroom","AI startup-incubator roast-sim: a five-persona 3D boardroom debating your idea.","Deterministic simulation where five distinct AI personas debate a startup idea in a 3D boardroom, one at a time, with human-quality voices and RAG-grounded citations from real sources.",[259,129,260,106],"Three.js","TTS","2026-07-15T10:55:07.805217+00:00","Founders rehearse pitches against friendly audiences and learn nothing about the objections that actually kill a round.","Five investor personas in a 3D boardroom, each grounded by RAG in a real corpus of their own writing and public positions, debating one at a time so the critique stays followable.","Running all five personas concurrently produced a wall of simultaneous text nobody could read. A paced director that gives one persona the floor at a time turned the same output into an actual conversation.","5","RAG-grounded personas citing real sources","archived","AI 창업 시뮬레이터. 5명의 투자자 페르소나가 3D 회의실에서 피치를 검증합니다.","창업자는 대개 우호적인 청중 앞에서 피치를 연습하기 때문에, 실제로 투자를 무산시키는 반론에 대해서는 아무것도 배우지 못합니다.","3D 회의실의 다섯 투자자 페르소나가 각자 실제 글과 공개 발언 코퍼스에 RAG로 근거를 두고, 한 번에 한 명씩 발언하도록 진행을 조율해 논의를 따라갈 수 있게 만들었습니다.","다섯 페르소나를 동시에 실행하니 아무도 읽을 수 없는 텍스트 더미가 나왔습니다. 한 번에 한 명에게만 발언권을 주는 진행자를 두자, 같은 출력이 실제로 따라갈 수 있는 대화가 되었습니다.","실제 출처를 인용하는 RAG 기반 페르소나",{"id":274,"title":275,"description":276,"long_description":277,"image":13,"image_url":13,"technologies":278,"category":281,"github_url":13,"demo_url":13,"featured":205,"status":56,"created_at":282,"updated_at":282,"problem":13,"approach":13,"trouble":13,"result_metric":13,"result_label":13,"role":13,"year":13,"tier":157,"link_status":267,"title_ko":13,"description_ko":283,"problem_ko":13,"approach_ko":13,"trouble_ko":13,"result_label_ko":13},"fdc3ad44-fd7c-4529-b804-c79d69d51530","Token Saver","Open-source RAG middleware that cuts token usage for Claude and ChatGPT.","Proxy plus browser extension that injects retrieval context to shrink prompts. Ships blob ingestion and an extension MVP; 123 tests.",[16,129,279,280],"Browser Extension","Proxy","Developer Tools","2026-07-14T10:55:07.805217+00:00","Claude와 ChatGPT의 토큰 사용량을 줄이는 오픈소스 RAG 미들웨어.",{"id":285,"title":286,"description":287,"long_description":288,"image":13,"image_url":13,"technologies":289,"category":82,"github_url":13,"demo_url":13,"featured":205,"status":56,"created_at":292,"updated_at":292,"problem":13,"approach":13,"trouble":13,"result_metric":13,"result_label":13,"role":13,"year":13,"tier":157,"link_status":267,"title_ko":13,"description_ko":293,"problem_ko":13,"approach_ko":13,"trouble_ko":13,"result_label_ko":13},"d8dc370c-45c0-4bf2-bc24-7c31cefc41ad","GeoLeads","General place-search web app over OpenStreetMap with real coordinates and area thumbnails.","Search any category of place (stadiums, pharmacies, hotels…) across a city via Overpass\u002FOSM, rendered on MapLibre with PostGIS and Carto area thumbnails.",[77,290,291,49],"MapLibre","Overpass\u002FOSM","2026-07-13T10:55:07.805217+00:00","OpenStreetMap 기반 장소 검색 웹 앱. 실제 좌표와 지역 썸네일을 함께 제공합니다.",{"id":295,"title":296,"description":297,"long_description":298,"image":13,"image_url":13,"technologies":299,"category":22,"github_url":13,"demo_url":13,"featured":205,"status":56,"created_at":301,"updated_at":301,"problem":13,"approach":13,"trouble":13,"result_metric":13,"result_label":13,"role":13,"year":13,"tier":157,"link_status":267,"title_ko":13,"description_ko":302,"problem_ko":13,"approach_ko":13,"trouble_ko":13,"result_label_ko":13},"8caf4c06-ef61-4860-8e92-d2ab39c02594","Rival Radar","Live competitive-intelligence app that streams an AI digest of a rival's web presence.","Crawls a competitor with Firecrawl, then streams an LLM digest with an adversarial verification pass to catch hallucinations.",[77,300,106],"Firecrawl","2026-07-12T10:55:07.805217+00:00","실시간 경쟁사 분석 앱. 경쟁사의 웹 활동 변화를 AI 요약으로 스트리밍합니다.",{"id":304,"title":305,"description":306,"long_description":307,"image":13,"image_url":308,"technologies":309,"category":54,"github_url":13,"demo_url":13,"featured":205,"status":56,"created_at":313,"updated_at":314,"problem":13,"approach":13,"trouble":13,"result_metric":13,"result_label":13,"role":13,"year":13,"tier":157,"link_status":267,"title_ko":13,"description_ko":315,"problem_ko":13,"approach_ko":13,"trouble_ko":13,"result_label_ko":13},"6df469d1-0f11-40c3-9ef5-1592f087d31c","Shorts Maker","Tri-lingual (UZ\u002FRU\u002FEN) pipeline that turns long YouTube videos into 9:16 shorts.","Uzbek-first shorts generator: transcription and edit-planning via Groq, reframing and captions with MediaPipe, rendered through Remotion. Multi-tenant from day one.","https:\u002F\u002Fatpyngridfrvfrbwkyyu.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fblog-images\u002Fproject-1784459937802-uvm9e725fx.png",[77,130,310,311,312,107],"RQ","MediaPipe","Remotion","2026-07-11T10:55:07.805217+00:00","2026-07-19T11:19:05.418+00:00","우즈베크어·러시아어·영어 3개 언어 파이프라인. 긴 YouTube 영상을 9:16 숏폼으로 변환합니다.",{"id":317,"title":318,"description":319,"long_description":320,"image":321,"image_url":321,"technologies":322,"category":82,"github_url":13,"demo_url":324,"featured":205,"status":25,"created_at":325,"updated_at":325,"problem":326,"approach":327,"trouble":328,"result_metric":329,"result_label":330,"role":32,"year":13,"tier":33,"link_status":34,"title_ko":13,"description_ko":331,"problem_ko":332,"approach_ko":333,"trouble_ko":334,"result_label_ko":335},"b23b2063-ba33-4bea-b800-9023e2d613f6","ChessCoach","Personal chess tutor with a Stockfish opponent and an AI coach that explains your moves.","Play against Stockfish WASM, then get natural-language coaching on your mistakes from an LLM analyst, with real Lichess piece art and Supabase-backed history.","https:\u002F\u002Fatpyngridfrvfrbwkyyu.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fproject-images\u002Fchesscoach.jpeg",[323,106,152],"Stockfish WASM","https:\u002F\u002Fxq-chesscoach.netlify.app","2026-07-10T10:55:07.805217+00:00","Stockfish 16 shipped a 38MB WASM binary to every visitor before the board could be used at all — unusable on a mobile connection.","Replaced the engine with a Stockfish 18 lite build carrying an embedded NNUE, behind a capability probe that runs multi-threaded where the browser allows it and falls back to single-threaded where it does not.","The obvious fix — lazy-loading the original binary — moved the stall rather than removing it. The payload itself had to change, and the lite build had to be verified to grade moves identically before it could replace the full engine.","7.3MB","down from 38MB, same playing strength","체스 학습·분석 웹 앱. 브라우저에서 Stockfish로 수를 평가하고 실시간 대국을 지원합니다.","Stockfish 16은 보드를 쓰기도 전에 모든 방문자에게 38MB WASM 바이너리를 내려보냈습니다. 모바일 환경에서는 사실상 사용할 수 없었습니다.","NNUE가 내장된 Stockfish 18 lite 빌드로 교체하고, 브라우저가 지원하면 멀티스레드로, 아니면 싱글스레드로 동작하도록 capability probe 뒤에 두었습니다.","가장 먼저 떠오르는 해법인 기존 바이너리 지연 로딩은 멈춤 지점을 옮길 뿐 없애지 못했습니다. 페이로드 자체를 바꿔야 했고, lite 빌드가 기존 엔진과 동일하게 수를 평가하는지 먼저 검증한 뒤에야 교체할 수 있었습니다.","38MB에서 줄임, 기력은 동일",{"id":337,"title":338,"description":339,"long_description":340,"image":341,"image_url":341,"technologies":342,"category":346,"github_url":13,"demo_url":347,"featured":205,"status":56,"created_at":348,"updated_at":348,"problem":13,"approach":13,"trouble":13,"result_metric":13,"result_label":13,"role":13,"year":13,"tier":157,"link_status":34,"title_ko":13,"description_ko":349,"problem_ko":13,"approach_ko":13,"trouble_ko":13,"result_label_ko":13},"562b744a-db9d-4110-9a22-2dbc20cf94a2","ScanDina","Barcode health-advisor and receipt tracker across web and Telegram.","Scan a product barcode for a health verdict, and track spending from receipts. Go + PocketBase backend with a web and Telegram frontend.","https:\u002F\u002Fatpyngridfrvfrbwkyyu.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fproject-images\u002Fscandina.jpeg",[343,344,345],"Go","PocketBase","Telegram","Mobile","https:\u002F\u002Fscandora-pitch.netlify.app","2026-07-09T10:55:07.805217+00:00","바코드 기반 건강 도우미 겸 영수증 관리. 웹과 텔레그램에서 모두 사용합니다.",{"id":351,"title":352,"description":353,"long_description":354,"image":13,"image_url":13,"technologies":355,"category":22,"github_url":13,"demo_url":13,"featured":205,"status":358,"created_at":359,"updated_at":359,"problem":13,"approach":13,"trouble":13,"result_metric":13,"result_label":13,"role":13,"year":13,"tier":157,"link_status":267,"title_ko":13,"description_ko":360,"problem_ko":13,"approach_ko":13,"trouble_ko":13,"result_label_ko":13},"b96f382e-6038-471e-b686-468205c97665","VoiceDesk","Cloned-voice AI receptionist for hotels, answering calls in a natural voice.","AI phone receptionist using a cloned voice for hotels, targeting Uzbek and Korean markets, built on a Vapi telephony spike.",[356,357,260],"Vapi","Telephony","In Progress","2026-07-08T10:55:07.805217+00:00","호텔용 음성 복제 AI 리셉셔니스트. 자연스러운 목소리로 전화를 응대합니다.",{"id":362,"title":363,"description":364,"long_description":365,"image":13,"image_url":13,"technologies":366,"category":82,"github_url":13,"demo_url":13,"featured":205,"status":358,"created_at":368,"updated_at":368,"problem":13,"approach":13,"trouble":13,"result_metric":13,"result_label":13,"role":13,"year":13,"tier":157,"link_status":267,"title_ko":13,"description_ko":369,"problem_ko":13,"approach_ko":13,"trouble_ko":13,"result_label_ko":13},"7c86f5b4-8607-410e-863b-e7dedc4bb35c","UniDesk","Unified inbox that pulls Telegram conversations into a web dashboard.","GramJS service feeding a FastAPI inbox backend and a Next.js inbox UI for managing many Telegram conversations in one place.",[367,130,77],"GramJS","2026-07-07T10:55:07.805217+00:00","통합 인박스. 텔레그램 대화를 웹 대시보드로 모아서 관리합니다.",{"id":371,"title":372,"description":373,"long_description":374,"image":13,"image_url":13,"technologies":375,"category":82,"github_url":13,"demo_url":13,"featured":205,"status":56,"created_at":377,"updated_at":377,"problem":13,"approach":13,"trouble":13,"result_metric":13,"result_label":13,"role":13,"year":13,"tier":157,"link_status":267,"title_ko":13,"description_ko":378,"problem_ko":13,"approach_ko":13,"trouble_ko":13,"result_label_ko":13},"f445dfbe-920f-4222-b124-edf6d3b7bfde","Hakgil","Korean university matcher that guides students from intake to a shortlist.","Three-screen flow (intake → results → detail) matching students to Korean universities, built from a design import.",[77,376],"Tailwind","2026-07-06T10:55:07.805217+00:00","한국 대학 매칭 서비스. 입력부터 최종 후보 목록까지 안내합니다.",{"id":380,"title":381,"description":382,"long_description":383,"image":13,"image_url":384,"technologies":385,"category":388,"github_url":13,"demo_url":389,"featured":205,"status":56,"created_at":390,"updated_at":391,"problem":13,"approach":13,"trouble":13,"result_metric":13,"result_label":13,"role":13,"year":13,"tier":157,"link_status":34,"title_ko":13,"description_ko":392,"problem_ko":13,"approach_ko":13,"trouble_ko":13,"result_label_ko":13},"2739e2bc-cfd0-459e-8eb4-bcd87afbd88d","BIGPIT","Cinematic storefront for a real Tashkent sports-supplements brand.","Nuxt 3 e-commerce experience with SVG scroll-cinematics, EN\u002FRU localization and a checkout flow.","https:\u002F\u002Fatpyngridfrvfrbwkyyu.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fblog-images\u002Fproject-1784460008962-hxejv3zojx.png",[386,153,387],"Nuxt 3","SVG","E-commerce","https:\u002F\u002Fbigpit-uz.netlify.app\u002F","2026-07-05T10:55:07.805217+00:00","2026-07-19T11:20:28.062+00:00","타슈켄트의 실제 스포츠 보충제 브랜드를 위한 시네마틱 스토어프론트.",{"id":394,"title":395,"description":396,"long_description":397,"image":13,"image_url":13,"technologies":398,"category":346,"github_url":13,"demo_url":13,"featured":205,"status":358,"created_at":401,"updated_at":401,"problem":13,"approach":13,"trouble":13,"result_metric":13,"result_label":13,"role":13,"year":13,"tier":157,"link_status":267,"title_ko":13,"description_ko":402,"problem_ko":13,"approach_ko":13,"trouble_ko":13,"result_label_ko":13},"3b5e4fcc-f483-43c7-8340-078722ef32f5","Pose Coach","Camera app that overlays an accurate human-outline pose suggestion on your scene.","Selfie-first camera app that suggests a flattering pose as an outline, using a Keras scene classifier to pick the pose for the background.",[399,311,400],"Keras","Camera","2026-07-04T10:55:07.805217+00:00","카메라 앱. 촬영 중인 장면 위에 적절한 인물 포즈 가이드라인을 겹쳐 보여 줍니다."]