Reports to:Chief Technology Officer
*Requires minimum of 2 days a week onsite in Boston, MA office
Base salary range: $135,000 to $170,000 USD, plus annual bonus. Final compensation will be determined based on experience, skills, qualifications, and relevant market data.
About the Opportunity
IANS is rapidly expanding its AI strategy toward a Data-as-a-Service (DaaS) model, delivering not only insights through our own applications, but also structured data feeds, APIs, and AI ready interfaces that enable clients to build their own intelligent systems and agentic workflows.
We are seeking an Agentic Engineer to build and ship the agent capabilities at the heart of IANS' agentic AI platform. This is a hands-on builder's role: working within the platform architecture set by our senior and principal engineers, you will implement, test, and operate the agents, tools, retrieval components, and evaluation suites that power both our internal AI products and the client-facing agent platform that consumes IANS data feeds. You will own well-scoped features end-to-end — from design through production — and grow your scope as you demonstrate ownership and judgment.
This is an exceptional opportunity to do serious agentic engineering early in your career: you will work daily with engineers who have shipped production multi-agent systems, on a platform where evaluation, observability, and security are first-class concerns rather than afterthoughts.
What You'll Do:
Agentic Systems Development
Build, test, and ship production-quality agent capabilities — tools, retrieval components, memory features, orchestration steps, and evaluation coverage — within the platform's established architecture.
Implement agent and tool-use patterns with frameworks such as LangChain, LangGraph, LlamaIndex, AutoGen, or comparable systems, with regression coverage and observability included in every change.
Integrate retrieval-augmented generation (RAG), structured tool use, MCP-style tool protocols, and APIs into robust, enterprise-grade platform components.
Contribute to the developer-facing primitives that allow external clients to safely extend the IANS agent platform with their own proprietary data and workflows.
Evaluation, Observability & Quality
Extend benchmarking, regression testing, and observability suites that measure agent quality, latency, cost, reliability, and safety, using modern AI observability tooling such as LangSmith, Langfuse, Arize, or Weights & Biases.
Write and maintain evaluations for agentic behaviors — tool-use correctness, hallucination rates, multi-turn coherence, and task completion — for the features you ship.
Investigate eval regressions and production incidents in agent behavior, and land the fixes.
Collaboration & Growth
Participate actively in design reviews and code reviews, absorbing and applying feedback from senior and principal engineers.
Take progressively larger ownership as you build a track record, from features to subsystems.
Stay current on the rapidly evolving agentic AI landscape (frontier models, orchestration frameworks, evaluation standards, agent protocols) and share what you learn with the team.
Software Engineering & Systems Integration
Ship production code across the stack: Go for services and agent runtimes, Python for AI/ML workflows, and React with TypeScript for customer-facing and internal web applications.
Build the APIs and microservices that internal teams and external clients use to integrate with IANS' data and agent platforms.
Test, log, trace, and performance-tune everything you ship, including agent workflows and model-driven systems.
AI Infrastructure & Deployment
Deploy and operate services on IANS' AWS-based agent infrastructure, including AWS ECS agent runtimes, AWS Bedrock for foundation model access, and AWS Lambda for tool execution.
Contribute to the data pipelines, vector databases, and retrieval systems that support RAG, agent memory, embeddings, and inference at scale.
Instrument token usage, latency, and inference cost for the features you own.
Security & Enterprise-Grade Standards
Follow IANS' standards for security, data isolation, governance, observability, and cost control in everything you ship, especially where agent capabilities and IANS data products are exposed to clients.
Implement access controls, sandboxing, and audit logging requirements in the components you build.
Build in compliance with regulatory frameworks (GDPR, CCPA, etc.) and IANS' SOC 2 Type II controls, and uphold responsible AI practices.
What You Bring:
Required
3–5 years of software engineering experience, including hands-on work building LLM powered features or agentic systems that reached production users.
Proficiency in Go, or strong proficiency in a comparable systems language (Rust, TypeScript, Java, C#) with the ability to become productive in Go quickly; working knowledge of Python for AI/ML workflows.
Experience building web applications with React (or a comparable modern frontend framework) and TypeScript.
Experience building and shipping LLM-powered applications — whether with an agent framework such as LangChain, LangGraph, LlamaIndex, or AutoGen, or directly against model APIs. Strong engineering fundamentals matter more to us than any particular.
Working understanding of RAG, embeddings, vector databases, and prompt/context.
Familiarity with cloud infrastructure, ideally AWS (ECS, Lambda, Bedrock, or comparable services).
Strong testing and debugging discipline, and the habit of instrumenting what you ship.
Clear written and verbal communication and a demonstrated appetite for feedback and growth.
Nice to Have
Exposure to AI evaluation and observability tooling such as LangSmith, Langfuse, Arize, or Weights & Biases.
Experience with MCP-style tool protocols or building tools for AI agents.
Experience operating services in production (on-call, incident response, performance tuning).
Familiarity with fine-tuning or post-training techniques (LoRA, PEFT, RLHF, DPO).
Contributions to open-source AI or agent tooling.
Cybersecurity domain familiarity.
Agentic engineering is a young discipline, and few candidates will check every box above. If this role excites you and you can show us strong engineering fundamentals and real work with LLM powered systems, we encourage you to apply even if you don't meet every single qualification— research shows that people from underrepresented groups often rule themselves out prematurely, and we'd rather make that call together.