We are building a B2B, multi-tenant portfolio intelligence platform for Australian financial advisers (ASX-focused). The system includes an agent/orchestration layer, model portfolios, risk profiling, tenure-based allocation (1–5 years), external market/news/fundamental APIs, and full audit/compliance logging. This is not a consumer trading app. It is an adviser-grade system with deterministic logic, explainability, and reproducibility. Scope of Work (Initial Phase) Build a backend system (FastAPI or Node/Nest) hosted on Replit or similar Implement agent/orchestration layer (ingestion, analytics, recommendation, monitoring) Integrate external APIs (ASX prices, fundamentals, news/sentiment – pluggable providers) Implement model portfolios, risk profiles, tenure-based allocation logic Design multi-tenant data model, audit logs, and versioned recommendations Expose clean REST APIs for adviser portal (UI can be basic initially) Required Skills Strong experience with Python (FastAPI) or Node.js (Nest/Express) Experience building agentic or event-driven systems (queues, workers, cron) Financial systems experience (portfolio logic, risk, market data) strongly preferred PostgreSQL, background jobs (Celery/BullMQ), API integrations Emphasis on deterministic logic, auditability, and versioning Nice to Have FinTech / Wealth / Adviser platforms experience Knowledge of ASX data providers Experience with PDF reporting, compliance workflows Frontend experience (React/Next.js or AI UI builders) Engagement Contract (20–30 hrs/week to start) Long-term potential if initial phase goes well Clear specs provided; founder is product- and tech-literate To Apply (Important) Please include: 1–2 relevant systems you’ve built (agentic, FinTech, data-heavy) Your preferred stack (Python or Node) and why Briefly explain how you would design an agent/orchestration layer Generic applications will be ignored. Please answer the following screening questions to qualify. 1. Describe how you would design an agent/orchestration layer that ingests market data daily, generates portfolio analytics, and produces deterministic recommendations. How do you handle retries and idempotency? 2. If an adviser generates a recommendation today and we need to reproduce it exactly 18 months later, what must be stored and versioned? 3. How would a 1-year vs 5-year investment horizon affect portfolio construction and risk constraints? 4. How would you design the system so that switching market data providers does not impact business logic or calculations? 5. What are common mistakes engineers make when building portfolio or risk systems without financial domain experience? 6. Given limited time and budget, what would you deliberately NOT build in Phase 1, and why?