Senior AI Engineer, TIFIN @Work

Tifin • Boulder, CO

9d • Onsite

About The Position

TIFIN @Work is an AI-native workplace platform that connects payroll, benefits, retirement plan, and outside account data with AI agents that provide personalized financial guidance to employees and useful signals to advisors. The Senior AI Engineer will own the architecture of the platform’s AI agents, designing multi-agent workflows that address financial and benefits questions, building systems to evaluate answer quality, and ensuring outputs—especially financial figures—can be traced to their sources. The role offers autonomy and close collaboration with product and client-facing teams. The team is small, distributed across the United States and Spain, and works in-office every day to collaborate. It ships in short cycles, conducts design reviews, tests thoroughly, and adapts its development practices as models evolve.

The engineer will build production workflows in which planner and specialist agents coordinate, use tools, stream responses, pause for human input, and recover from checkpoints. They will develop a quality evaluation system using adversarial tests, execution-trace scoring, calibrated LLM judges, CI regression gates, production monitoring, and a published quality score. They will also build a retrieval system that turns financial and benefits documents into grounded, cited answers with an evidence trail, and a personalization context layer that controls what agents know and may use while improving security, latency, and cost.

The role requires production experience with agentic AI, orchestration frameworks, evaluation and observability, machine learning, retrieval-augmented generation, model serving, and full-stack/cloud engineering. The engineer is expected to make thoughtful trade-offs, communicate decisions in writing and design reviews, review teammates’ designs and merge requests, and explore emerging models, agent patterns, and AI-native development tools. Compensation is $150,000–$200,000 USD, with meaningful equity and eligibility for a benefits program.

Requirements

• Experience shipping agentic AI to real users.
• Deep understanding of orchestration, state, checkpointing, and streaming, with substantial experience using LangGraph or a similar framework.
• Experience building evaluation systems, including benchmarks and golden datasets, execution-trace scoring, LLM-as-a-judge calibrated against human labels, and offline and online evaluations such as A/B tests.
• Experience using tracing and observability tools such as LangSmith, Phoenix, or Langfuse.
• Machine learning experience fine-tuning or distilling small language models for narrow tasks such as classification, routing, or extraction, using techniques such as LoRA, QLoRA, and quantization.
• Experience serving models with inference stacks such as vLLM, TGI, or SGLang and weighing accuracy against latency and cost.
• Experience building production RAG systems using chunking, hybrid search, reranking, vector stores, grounding, and citations.
• Experience making tool calls and structured outputs reliable with Pydantic or JSON Schema.
• Understanding of PII protection before data is sent to an external model.
• Experience shipping with more than one model provider and selecting providers based on quality, speed, cost, and reliability.
• Strong Python skills and comfort working across FastAPI or Django, React, TypeScript, data infrastructure, and cloud.
• Ability to explain decisions clearly in writing and design reviews and to review teammates’ designs and merge requests rigorously.
• Interest in building AI systems that help people make better financial decisions.

Nice To Haves

• Interest in fine-tuning and serving open-weight models.
• Interest in MCP and tool-protocol design.
• Interest in generative UI.
• Experience using AI-native development tools such as Claude Code, Cursor, or Codex.

Responsibilities

• Design and build production multi-agent workflows in which planner and specialist agents coordinate, call tools, stream responses, pause for human input, and recover reliably from checkpoints.
• Create an evaluation system that includes adversarial test sets, execution-trace scoring, calibrated LLM judges, CI regression gates, production monitoring, and a published quality score.
• Build a retrieval system that produces grounded, cited answers from financial and benefits documents using hybrid search, reranking, and thoughtful chunking, with an evidence trail behind every number.
• Build the context layer that determines what each agent knows and may use.
• Improve system security, latency, and cost through scoped permissions, validation, prompt caching, and intelligent model routing.
• Make tool calls and structured outputs reliable, and treat prompts and tool contracts as versioned and tested artifacts.
• Use tracing and observability to connect production outcomes to development and establish feedback loops for improvement.
• Make model-provider choices based on quality, speed, cost, and reliability.
• Work directly with product and client-facing teams and exercise autonomy over what gets built.
• Communicate decisions clearly in writing and design reviews, and review teammates’ designs and merge requests.
• Experiment with new models, agent patterns, and AI-native development tools, and translate promising capabilities into dependable products and faster workflows.

Benefits

• Meaningful equity stake as a significant part of the overall compensation package.
• Benefits program eligibility, including 90–100% medical, vision, and dental coverage for the employee and dependents.
• Retirement benefits.
• Flexible PTO.

What This Job Offers

• Job Type: Full-time
• Career Level: Senior
• Education Level: No Education Listed