Note: The job is a remote job and is open to candidates in USA. Allio Capital is building an intelligence platform for financial advisors and fintechs focused on delivering governed AI solutions. The Financial AI Engineer will design and implement AI workflows and orchestration that enhance the advisory experience by ensuring data-driven and compliant outputs.
Responsibilities
- Build the AI advisory layer end to end — design and ship retrieval-augmented generation (RAG) over market data, portfolio state, methodology, and compliance documentation, so the system answers with grounded, citable evidence rather than plausible-sounding guesses
- Design agentic systems that do real work — build agentic AI workflows that plan, call tools, query our quant services, and compose multi-step answers across portfolio rebalancing, multi-account optimization, and scenario analysis
- Own the orchestration layer — architect the AI orchestration that routes between models, tools, retrieval, and our quantitative services (via LangChain + Amazon Bedrock, with MCP for tool/data access), balancing latency, cost, and correctness
- Connect AI to the quant engine — wire the intelligence layer into portfolio optimization, risk modeling, and tax optimization (tax-aware rebalancing, lot-level / tax-loss-harvesting logic), where small, well-reasoned changes have large downstream impact
- Productionize ML and data pipelines — build reproducible, testable pipelines for retrieval, feature generation, regime modeling, and walk-forward validation using Kedro (or similar) so experiments become dependable production systems
- Ship across the stack — contribute to NestJS /GraphQL backends, FastAPI Python services, and the React /TypeScript advisor experience; you move fluidly across them rather than staying in one lane
- Build for governance and trust — implement evaluation harnesses, guardrails, grounding/citation, and audit trails so AI output is measurable, defensible, and aligned with SEC Marketing Rule and AI-disclosure expectations
- Operate in production — AWS (EKS, Bedrock, RDS, S3, SNS/SQS), event-driven services, observability; you debug and harden your own work, not just author it
Skills
- 5+ years building and shipping production software, with 3+ years working hands-on with LLM-based or ML systems
- Real RAG and LLM-application depth — you've built production retrieval pipelines (embeddings, vector search, chunking/grounding strategies, evaluation) and understand why they fail, not just how to wire them up
- Agentic and orchestration experience — you've designed multi-step, tool-using agents and the orchestration around them (LangChain, Amazon Bedrock, or comparable; MCP a strong plus)
- Strong full-stack foundation — modern TypeScript (Node/NestJS or similar; GraphQL a plus), Python (FastAPI or similar), and React
- Quantitative / financial-modeling aptitude — comfort with portfolio math, statistics, or numerical Python (NumPy/pandas) and the judgment to reason about correctness where there's no compiler to catch a bad assumption. Direct experience with portfolio optimization, tax optimization, risk modeling, or back-testing is a strong plus
- Pipeline and MLOps maturity — experience making ML/data pipelines reproducible and production-grade (Kedro, walk-forward validation, or comparable tooling)
- Production ownership mindset — you've operated services in the cloud (AWS preferred), written meaningful tests, and take responsibility for the quality of what you ship — especially in money-handling and financial-modeling code, where we hold a high bar for correctness
- LLM evaluation and observability tooling (offline evals, tracing, regression suites for non-deterministic systems)
- Comfort with ambiguity and autonomy — you can take a loosely-specified, high-value problem, lock the scope, and deliver
- Clear communicator who works well in a small, hyper fast moving, senior, remote team
- Fintech, wealthtech, or RIA/broker-dealer domain experience
- Familiarity with compliance-sensitive AI — evals, guardrails, model governance, or building under SEC Marketing Rule / AI-washing constraints
- ML/MLOps exposure (regime modeling, walk-forward validation, pipeline tooling)
- Microservices, event-driven architecture (SNS/SQS), MikroORM/PostgreSQL, Kubernetes/Terraform
Benefits
- Genuine ownership — senior team, no bureaucracy, you ship to production quickly.
- Hard, meaningful problems — quantitative finance + AI, where engineering quality •is• the product.
- Governance as a moat — we're building AI that's trustworthy by design, not bolted on, in one of the most compliance-sensitive domains there is.
- High leverage — your work lands in front of real advisors and API partners, not in a backlog.
- Remote-first.
Company Overview