AI Engineer (TIFIN.ai)

• Our north star is delivering better wealth outcomes for more people. We’re building production-grade multi-agent systems that power advisor copilots, investment intelligence workflows, and autonomous research capabilities
• Our AI Engineers architect, build, and operationalize these systems at scale, pushing the boundaries of what agentic AI can do. We’re hands-on engineers focused on shipping reliable, enterprise-ready agentic AI systems into production
• Design and implement production-grade multi-agent systems and ReAct-style reasoning loops using modern agent frameworks and orchestration engines (e.g., LangGraph, Pydantic AI, Agent Harness, Tool-Calling, Code Execution)
• Build agent workflows that integrate RAG-based retrieval, agent memory, and knowledge graphs for grounded, long-horizon reasoning, fine-tuning models where prompting and retrieval alone fall short
• Establish evaluation and benchmarking frameworks, including LLM-as-judge pipelines, for multi-step reasoning accuracy, groundedness, hallucination mitigation, and financial correctness
• Design and enforce AI governance — audit trails, guardrails, and human-in-the-loop checkpoints — appropriate to a regulated industry
• Develop distributed agent services with strong observability and failure handling, and optimize latency, cost, and infrastructure decisions across model serving, vector/graph databases, and caching
• PROJECTS:
• Agent Advisor Copilot. Build a production-grade, multi-agent copilot for financial advisors that retrieves and reasons over client data using RAG and a persistent knowledge graph of clients and holdings, analyzes portfolio exposures and risk scenarios, generates personalized insights, enforces compliance guardrails, and drafts client-ready communications — all within a monitored, auditable architecture graded by automated evals and LLM-as-judge review
• Deep Research & Workflow Agents. Design end-to-end AI workflows spanning client discovery, investment research synthesis, portfolio construction and optimization, and compliant meeting preparation — powered by deep research agents that reason over live web and internal data in ReAct-style loops — replacing fragmented tools with intelligent, autonomous systems
• Agentic Infrastructure & Reasoning Stack. Architect a scalable multi-agent platform on LangGraph-style orchestration, with agent memory and state management, dynamic tool invocation, model fine-tuning pipelines, structured output validation, observability, fault tolerance, and automated evaluation — solving reliability, explainability, and regulatory challenges at scale

Benefits

• Performance linked variable compensation, including equity
• Medical, dental, vision, life and disability insurance
• 401k Retirement Account through Empower Retirement
• Flexible PTO policy and Company-paid holidays
• Corporate Social Responsibility and volunteering opportunities
• Fully stocked office kitchens
• Access to our Chief Mindfulness Officer, including 1:1 personal coaching for executives and rising stars
• Company sponsored events like Qi Gong, Lunch & Learns, Development Workshops, Dinners, Happy Hours, and more!
• Parental leave: 12 week paid maternity, 4 week paid paternity leave

• Strong systems thinking, an ability to design beyond single-model solutions toward coordinated, multi-component architectures
• Deep proficiency in Python, with experience in distributed systems, cloud infrastructure (AWS/GCP/Azure), and containerized deployments
• Strong experience with RAG, agent memory, knowledge graphs and graph databases, and LLM fine-tuning
• Resilience and adaptability, driven by a desire to build state-of-the-art AI that drives better wealth outcomes for more people
• Experience implementing evaluation frameworks, benchmarking, and LLM-as-judge pipelines, along with monitoring and reliability safeguards for AI systems
• 3+ years of experience building and shipping Generative AI and LLM applications into production, with demonstrated experience designing and deploying multi-agent systems, including LangGraph or comparable orchestration frameworks and ReAct-style deep-agent reasoning loops
• Candidates must be authorized to work in the US without sponsorship