About

Project: Build AI-Powered Matching MVP (OpenAI + Pinecone + Google Forms) We are building an MVP for SecondActPro, a new freelance marketplace that connects experienced/retired professionals with organizations seeking trusted, flexible talent. The MVP should demonstrate a working pipeline where professionals and organizations submit structured data, embeddings are created and stored, and the system returns relevant matches with a short “explanation.” Scope of Work 1. Intake Forms (Google Forms) Two forms: one for professionals, one for organizations. Fields should include: Professionals: name, background/skills, years of experience, preferred roles, availability, “do not want” exclusions. Organizations: project type, required skills, duration, budget range, “do not want” exclusions. Smart branching (earlier choices change later questions). Must output structured data compatible with embedding pipeline. 2. Embeddings Use OpenAI text-embedding-3-small (cheaper, good for MVP). Each new form submission is automatically embedded via Google Apps Script → OpenAI API. Store embedding + metadata. 3. Vector Database Use Pinecone (preferred) or alternative like Qdrant or PostgreSQL with pgvector. Create indexes for both professionals and organizations. Implement basic similarity search. 4. Matching Logic Simple vector similarity (top-N matches). Include support for optional filters (availability, project type, exclusions). Provide results ranked by match score. 5. Lightweight API Build small API (FastAPI or Flask) with endpoints: /embed – embed and return vectors. /upsert – save new entries. /match – return top candidates or projects. 6. Explanation Layer (LLM) Use ChatGPT API (gpt-4o-mini or gpt-4o) to generate a short human-readable reason for match. Example: “We matched you with Jane because she has skills in program management and healthcare compliance, which aligns with your project on…” 7. Testing & Pilot Populate with sample data (10–20 entries). Run test matches and tweak filters/weights for balance. Deliver documented setup and instructions. Deliverables Working Google Forms (2). Configured Pinecone DB + API keys set up. Scripts (Google Apps Script + Python) for embedding and upserting. API endpoints (FastAPI). Sample test runs + documentation. Ideal Candidate Experience with Python, OpenAI API, embeddings, Pinecone/Qdrant/Postgres, FastAPI/Flask, Google Apps Script. Can deliver a clear, documented MVP that we can extend later. Strong communicator who can recommend efficient shortcuts. Nice-to-Haves (Optional but Valuable) Prior experience building matching/recommendation engines. Familiarity with marketplaces, talent platforms, or job boards. UX/UI knowledge (to eventually improve forms and user journey). Ability to advise on cost optimization (e.g., API usage, database scaling). Understanding of data privacy & compliance (since professionals will share personal info). Budget & Timeline This is an MVP, not a polished production build. Open to proposals on budget, but we expect a lean and efficient implementation (1–2 weeks). Longer-term work possible if MVP is successful. Why this matters: SecondActPro helps experienced professionals find flexible, meaningful work while enabling organizations to access wisdom and impact without full-time commitments. Your work will be the technical backbone of this mission.