Description
Summary Project overview We are looking for an experienced Discord and AI automation engineer to build several connected bots for the Fifth Percent engineering community. Fifth Percent runs technical workshops and a Discord community for AI, ML, software, data, and platform engineers. Our focus is production AI: reliability, RAG, evaluation, observability, infrastructure, latency, cost, and real failure modes. This is not a basic moderation bot or a simple ChatGPT wrapper. We need a maintainable system that improves content, participation, events, and community operations without making the Discord feel automated. Bots we want to build: 1. Automated technical content bot - Search approved websites, engineering blogs, research papers, GitHub, RSS feeds, and technical communities. - Score content for relevance, credibility, depth, originality, and discussion potential. - Detect duplicate topics. - Createnatural, conversational posts for the correct Discord channels. - Link to the original source and ask specific technical questions. - Respond selectively when it can add genuine value. Posts should feel like useful community contributions, not automated news summaries or an assistant speaking to members. 2. Conversation and engagement bot - Identify worthwhile discussions that have slowed down. - Surface unanswered questions. - Suggest useful follow-up questions, previous threads, or relevant sources. - Avoid interrupting active conversations or replying to every message. 3. Technical Q&A bot - Use a /ask workflow to help members submit clear technical questions. - Collect stack, scale, expected behaviour, actual behaviour, constraints, logs, and attempted fixes. - Create structured help threads. - Track unanswered and resolved questions. - Route questions to opted-in members with relevant expertise. 4. Member onboarding and matching bot - Collect members' technical interests, expertise, projects, time zones, and learning goals. - Recommend relevant channels and discussions. - Match members for architecture reviews, repository reviews, workshops, and collaborations. 5. Workshop and events bot - Share setup instructions, repositories, schedules, and event resources. - Collect attendee questions and create event-specific threads. - Run technical polls and capture unanswered questions. -Send follow-up exercises and recommend relevant workshops or community discussions. 6. Community insights bot - Identify recurring technical topics and community needs. - Create weekly community digests. - Capture potential workshop ideas, articles, social posts, FAQs, and event topics. - Store approved insights in a searchable internal content bank. Nothing should be published externally without human approval. Moderator dashboard Our team should be able to: - Configure channels, topics, sources, and posting frequency. - Set quiet hours, response limits, and approval rules. - Pause individual bots or the entire system. - Review scheduled posts, source links, activity logs, and error logs. - Block sources or topics and manage role permissions. - Search community insights and view performance metrics. Required experience You should have strong experience with: - Discord API, Discord.js, discord.py, or equivalent. - Python or TypeScript/Node.js. - LLM APIs, tool calling, and retrieval-augmented generation. - Web search, RSS, GitHub API, and webpage extraction. - PostgreSQL, pgvector, or another vector database. - Embeddings, semantic search, and duplicate detection. - Background jobs, scheduling, and conversation-state management. - Admin dashboards, cloud deployment, monitoring, testing, and cost controls. - Prompt-injection protection and secure credential management. The system must use Discord's supported bot API. We will not consider self-bots, personal user-token automation, or browser automation controlling normal Discord accounts. Initial delivery We expect the project to begin with a paid discovery and architecture phase, followed by a milestone-based build. The first MVP should prioritise: 1. Automated technical content posting. 2. Conversation and engagement functionality. 3. Technical Q&A and unanswered-question tracking. 4. A basic moderator dashboard. 5. Source controls, duplicate detection, logging, and emergency pause. The remaining bots can be delivered in later phases. Engineering expectations - Modular, maintainable architecture rather than a fragile collection of prompts. - Secure credential management, development and production environments, automated tests, structured logs, and retry handling. - Provider abstraction where practical, allowing future changes to the LLM, search provider, or vector store. - Clear deployment documentation, technical handover, and realistic estimates for monthly model, search, hosting, and database costs. Application questions Please include: 1. Examples of relevant Discord or AI automation systems you have built. 2. Your recommended architecture and technology stack. 3. How you would make automated posts feel natural and non-repetitive. 4. How you would verify web sources and reduce unsupported claims. 5. How you would prevent excessive posting and bot domination. 6. How you would protect against prompt injection. 7. Your proposed timeline, milestones, and estimated cost. 8. Expected monthly hosting, model, search, and database costs. 9. Your availability for ongoing support. Please begin your proposal with: "Production is the real exam." Commercial structure We are open to a paid discovery phase followed by a fixed-price build, a milestone-based fixed-price project, or an hourly engagement with clear weekly deliverables. Please include your rate or estimate, proposed timeline, milestone breakdown, expected monthly running costs, relevant work examples, and availability for maintenance. Final objective The goal is to build bots that make the Discord more useful, better informed, and easier to participate in - without making the community feel automated.