Project: Relei — AI project-to-content automation for consultancies
Type: Freelance, full-time (≈35–40h/week, 4–6 weeks, milestone-based)
Location: Remote (CET overlap preferred)
Start: Immediate
About Relei
Relei helps consultancies transform their project archives into reusable business development content — from case studies to proposals and points of view.
We’re working with pilot customers (e.g. design and innovation consultancies) to automate how teams turn their project knowledge into sales-ready material.
The Role
You’ll take over and extend an existing prototype to make it pilot-ready by year-end.
That includes refining the PoV Agent workflow (extracting insights, generating structured content), improving prompt orchestration, and preparing the system for live pilot use.
You’ll work directly with the founder and occasional technical advisors — a mix of autonomy and collaboration, focused on tangible milestones.
Responsibilities
- Review current prototype and architecture (Python + FastAPI + AWS)
- Improve LLM orchestration (LangChain / custom pipelines) and data flow
- Implement structured JSON-based content outputs for PoVs and proposals
- Integrate with AWS stack (Lambda, S3, Secrets Manager) and handle deployment
- Support simple UI integration or API endpoints for pilot testing
- Write clear documentation for handover and iteration
Tech Stack
- Core: Python, FastAPI, LangChain, OpenAI API
- Infra: AWS (Lambda / S3 / Secrets Manager), GitHub
- Optional: RAG workflows, Perplexity/Tavily integration, lightweight React front-end
You’re a fit if you:
- Have built production-level LLM applications (not just demos)
- Can independently scope and deliver under startup-style ambiguity
- Value pragmatic, milestone-driven work and async communication
- Enjoy solving messy early-stage problems and seeing impact quickly
Engagement
- Full-time for 4–6 weeks (extendable)
- Milestone-based delivery (e.g. Walking Skeleton → Internal Review → Pilot Demo)
- Paid 10-hour test task before full engagement
- Competitive rate depending on experience