See how candidates think and approach the work this role demands, before the phone screen. We'll build a video challenge from this posting, and you can edit or share it before it goes live.
Hebbia is building a specialized frontend layer for agentic AI systems, signaling that agent UX—not just agent capability—is becoming a distinct, defensible product surface that requires dedicated engineering focus.
The user interface for universal knowledge capture. We want every atom of information that our users need to be at their fingertips.
Designed to be generally capable– it can tackle even the most complex tasks, citing answers over any amount of sources. By showing its work, Hebbia empowers users to collaborate with AI on each step and validate responses instead of blindly trusting them. Our mission is to put capable AI in the hands of 1 billion people by 2030.
Hebbia enables leading finance and legal firms with advanced reasoning, copiloting and retrieval capabilities - unlocking meaningful insights for meaningful use cases. The Agents team builds everything from core document understanding capabilities to co-piloting experiences for matrix and deep, multi-source research. We’ve built our own agentic frameworks powered by distributed systems built for scale.
We don’t build for one-off success. We build steerable, reliable and explainable agentic systems. And we build these systems for the scale of data our customers bring to the table. Our goal is to unlock the unknowable unknown for customers all over the world.
We want to build a product that becomes indisposable to our customers and as delightful as your favorite consumer product. We move fast and build first of their kind systems.
Our team works in person 5 days a week at our SoHo office in NYC.
What we’ve built:
A custom multi-agent framework powering everything from deep research capabilities to copiloting interfaces paired with distributed systems infrastructure for long running agentic tasks.
The world’s most powerful and scalable LLM inference engine - a distributed, asynchronous DAG orchestrator capable of incorporating live graph mutations, cooperating in tandem with our LLM throughput management capabilities.
Elastically scaling data representation and metadata generation → powering the most effective private data retrieval systems.
Industry leading agents solving problems from buy side company diligence to multi billion dollar M&A.
As a Front-End Engineer on Agents, you will not only own the interfaces bankers, investors, and lawyers use to work with Max and Matrix, but the user experience overall. You will be responsible for simple, delightful experiences for users to direct long-running agents, review citations, and iterate on key deliverables. You will turn new agentic capabilities from our research team into fast, polished interfaces on the web and on iOS. Engineers at Hebbia are given agency to own features end-to-end, so when a feature needs a new endpoint, streaming API, or Hebbia MCP tool, you will build it too.
This role requires deep expertise in modern web front-end engineering (TypeScript, React), strong product sense, and comfort working in the backend when needed. iOS experience is highly valued: we are bringing Max and more to mobile, and you would help build it. You will work directly with applied researchers and infrastructure engineers, so you should be comfortable understanding the full system, from model call to UI.
Our team works in person 5 days a week at our SoHo office in NYC.
Design and build the agent-facing UI for Hebbia: streaming agent output, real-time progress over long-running tasks, citation and provenance views, checkpoint and approval flows.
Own front-end architecture for data-dense, real-time surfaces — hundreds of companies × thousands of documents — with rendering performance and state management that hold up at that scale.
Contribute to making mobile, a high-value surface for finance professionals, an excellent surface to talk to Max on the go.
Build the backend surface those UIs depend on: streaming and real-time endpoints, agent state and session APIs, Hebbia MCP tools, and the glue to new data integrations.
Partner with applied researchers to get new agentic capabilities in front of users quickly, then iterate using insights from eval frameworks.
Own quality end to end: accessibility, telemetry, testing, performance budgets, and a design system that scales across web and mobile.
Bachelor's degree in Computer Science, Engineering, or related field, or equivalent experience
4+ years building production front-ends at a venture-backed startup or top technology firm
Expert in TypeScript and React, with a deep understanding of rendering performance, state management, data fetching, and real-time patterns.
Comfortable across the stack — designing APIs, working with async services, and debugging the systems behind your UI
Strong product sense and design taste; comfortable working closely with designers and product managers.
Frequent user of AI products, especially during the development lifecycle (i.e. Cursor, Claude Code, etc.)
Excellent problem-solving, communication, and teamwork abilities
Bonuses:
You've shipped native mobile apps and understand the App Store release cycle end to end
Experience building LLM- or agent-facing products
The salary range for this role is $160,000 to $300,000, covering levels from junior to principal. Final leveling is determined through our assessment process, and exceptions to this range may be made for candidates with qualifications that fall outside our standard framework.
Life @ Hebbia
PTO: Unlimited
Insurance: Medical + Dental + Vision + 401K
Eats: Catered lunch daily + doordash dinner credit
Parental leave policy: 3 months non-birthing parent, 4 months for birthing parent
Fertility benefits: $15k lifetime benefit
New hire equity grant: competitive equity package with unmatched upside potential
#LI-Onsite
Answer a McCoy challenge. Companies hiring in engineering see your response and reach out when they want to talk.
About 15 minutes to prepare and 5 minutes to record. Unlimited retakes and a teleprompter help you answer. No AI training, and you can delete your response anytime.
This isn't an application to Hebbia.
hebbia:038bddb3-c0dd-43e1-8863-17efdf9427dbhebbia-aiThis posting hasn't been probed by our closure verifier yet. Stream C runs on a rolling schedule against postings approaching the close-decision threshold.
See how we measure for definitions, or our corrections log for known issues. Found something wrong? Flag a correction.