도구/프로젝트

벡터 DB 오픈소스 ChromaDB가 1.0을 발표했습니다. 필터링 성능 개선과 인덱스 안정성이 핵심이라고 하네요. 기존 0.4에서 마이그레이션 해보신 분 계신가요?

둘 다 써봤는데, Open-webui는 Ollama 연동이 편하고 LibreChat는 멀티모델 관리가 좋았습니다. RAG 기능은 LibreChat가 좀 더 성숙한 느낌? 여러분 선택은?

프롬프트/비용/지연을 한 대시보드에서 보니 운영 난이도가 확실히 내려갔습니다. 셀프호스팅도 가능해서 사내 도입에 부담이 없네요.

로컬 RAG 서비스 최소 구성 템플릿을 만들어봤습니다. docker compose 하나로 Ollama + FastAPI + ChromaDB가 뜹니다. 피드백 환영해요.

이메일 요약 → 슬랙 전송 → 주간 리포트 생성까지 n8n으로 묶어봤습니다. 노코드인데 LLM 노드가 잘 돼 있어서 생각보다 강력하네요.

폐쇄망용 완전 오프라인 RAG. Ollama만으로 동작, 임베딩은 bge-m3. 피드백 환영합니다.

Updated list of open LLMs you can use commercially. Handy when picking a model for a client project.

네임스페이스 API가 안정화됐고 성능도 올라갔다고 합니다. 마이그레이션 가이드 공유합니다.

비용/지연/프롬프트 버전까지 한 번에 보이니까 디버깅 시간이 반으로 줄었습니다. 무료 티어로 충분해요.

개발자용 월간 AI 뉴스 큐레이션 레포입니다. 매월 최신 LLM/도구 정리돼 있어서 스크랩해두면 좋아요.

We went LibreChat for multi-user + RBAC. Open-webui has nicer UX though. Your pick?

Keep this tab open. Everything from o1 to GPT-5-Codex-Mini lands here first.

Ran a quick benchmark. pgvector with HNSW was 80% the speed at 10% the cost. Details in thread.

Feed it a CSV export, it finds anomalies and writes a Korean explanation via an LLM. Useful for MSP folks.

One endpoint for all providers, consistent interface, and the config is plain YAML. We swapped out three vendor SDKs in an afternoon. If you're multi-provider, just use it.

드래그앤드롭으로 LLM + 검색 + 슬랙 연동까지. 개발자도 가끔은 로우코드가 편하네요.

I packaged my go-to RAG setup into a template. Docker compose up and you have a working API in 2 minutes.

듀얼 3090으로 70B 돌리려다 이틀 삽질. 결국은 OLLAMA_NUM_GPU + 스플릿 모드 조합으로 해결. 정리해봤습니다.

지금은 Git으로 관리하는데, 버전별 A/B 테스트가 필요해졌습니다. 전용 도구 추천 부탁해요.

CSV of prompts → JSONL of responses + scores. 100 lines of Go. Maybe useful for your eval pipeline.

Gradio got multimodal chat components. Streamlit is catching up. For internal demos, which wins?

We were merging PRs blind. Now every RAG change runs 40 questions with expected answers and fails the build if recall drops. It's crude but it caught two regressions last week. Sometimes the simple stuff wins.

50 lines of TypeScript, points at our search API, and now everyone queries docs from their editor. It's dumb but it changed how we work. The protocol is clunky but the integration win is real.

LangSmith is smoother but the per-seat pricing hurt us. Langfuse self-hosted costs us one small VM. Missing a few features but the trace UI is good enough and the data stays on our infra. For EU clients that matters.

Been running a 70B Q4 on one 80GB card for a side project. Key tricks: prefix caching on, max_num_seqs tuned, and a tiny health check for the load balancer. Sharing my config because I wish someone had shared theirs.

Ran the same 1M-doc dataset through all three. Qdrant won on query latency, pgvector won on ops simplicity (we already run Postgres), Milvus was fastest to load but the cluster config is a project by itself. Full numbers in the post.

We needed retries, state, and human-in-the-loop checkpoints. n8n got us to demo in a day, Temporal was a week of learning but the durability is worth it. For anything customer-facing we now default to Temporal.

Wrote our own gateway instead of buying one. Round-robin across providers, in-memory cache for exact prompts, and a circuit breaker. It's not fancy but it cut our API bill 35% and we control everything.

We were copy-pasting prompts into git and it was chaos. This tool stores prompts as versioned files, diffs them, and can push to any provider. Early days but it's been useful for our team of 4.

Every prompt change gets a shadow run: same input, old and new config, human review on a sample. Catches regressions the evals miss. Takes one extra table and a cron job. Best $0 investment we've made this quarter.