LLM Engineer / Agentic Prototyper (AI-Native)
Prototype the intelligence layer that powers Skiffra’s agentic orchestration platform.
About Skiffra
Skiffra is building the operating system for intelligent enterprises. We connect people, data, and AI agents into one adaptive fabric that learns continuously and moves with purpose. Across industries, organizations are slowed by fragmented systems, unshared expertise, and workflows that bury insight. Skiffra replaces that waste with orchestration that amplifies human judgment through fast, adaptive intelligence.
We’re not automating yesterday’s tasks. We’re designing the infrastructure for the intelligent enterprises of tomorrow—where humans and agents collaborate through fluid workflows, shared context, and real-time reasoning.
The Role
We’re hiring an LLM Engineer / Agentic Prototyper to architect and prototype the intelligence layer at the center of Skiffra’s platform. You’ll build, tune, and orchestrate AI agents that reason, collaborate, and act across complex enterprise environments. Your work will shape how intelligence flows across Skiffra—defining the early patterns for adaptive workflows, agent collaboration, and human-in-the-loop control.
This is an execution-first role. You’ll prototype agents that operate in real environments, handle messy inputs, survive failure modes, and deliver value inside the constraints of industrial and enterprise workflows. You’ll work across model behavior, tool design, reasoning loops, safety constraints, observability, and workflow orchestration—turning ambiguous problems into elegant agent systems that learn and improve over time.
What You’ll Do
Prototype agent teams that perform real tasks across finance, operations, and cross-functional workflows
Design runtime guardrails, validators, and safety constraints for predictable behavior under uncertainty
Build tools and interfaces that agents use to perceive data, trigger actions, and collaborate with humans
Evaluate model behavior using structured tests, adversarial prompts, edge cases, and runtime telemetry
Implement structured memory systems (episodic, semantic, vector-based) that support multi-step reasoning
Add observability hooks to track reasoning steps, tool calls, errors, and decision traces
Experiment with architectures for multi-agent collaboration, model selection, long-horizon planning, and adaptive loops
Work closely with product, data, and full-stack engineering teams to align agents with real workflows
Document patterns, discoveries, limits, and breakthrough behaviors for future automation and reuse
Shape Skiffra’s agentic engineering culture—hands-on, explorative, and disciplined
You’ll Thrive Here If You
Treat LLMs as reasoning systems, not autocomplete machines
Think in loops, flows, constraints, and failure modes
Care about agents that actually do work—not demos, but field-ready systems
See model quirks as entry points for deeper understanding
Build fast, learn fast, and experiment with urgency
Prefer prototyping over theorizing
Thrive in ambiguous problem spaces where clarity must be invented
Believe enterprise intelligence will be shaped by adaptive, agentic collaboration
What You Bring
Strong engineering foundation in ML engineering, software engineering, systems design, or related fields
Experience with LLMs, agent frameworks, prompt engineering, or autonomous workflows
Ability to design, debug, and evaluate multi-step reasoning chains and decision loops
Understanding of model failure modes and how to constrain them with tools, rules, and observability
Experience collaborating across product, data, and engineering to ship end-to-end prototypes
Demonstrated ability to learn at extreme velocity and adapt to new frameworks and capabilities
Comfort in hybrid roles that span research, prototyping, architecture, and engineering
Curiosity, humility, and a bias toward documentation and teachability
Why Skiffra
You’ll join a small founding team working to eliminate the inefficiency of how organizations operate today. You’ll prototype the intelligence layer that removes that drag and amplifies human capability. At Skiffra, your ideas won’t sit in docs—they’ll be deployed into operating mines, manufacturing floors, and enterprise environments within months.
You’ll have ownership over architecture decisions, research direction, and the chance to define new patterns for human–AI collaboration. This is a builder’s seat at the frontier of enterprise intelligence.
Details
Location: Hybrid – Los Angeles preferred, with flexible remote and on-site sprints
Type: Full-time, founding team
Compensation: Competitive salary + meaningful equity
Benefits & Eligibility:
This position is eligible for Skiffra’s standard benefits, including health insurance and retirement plans, and may also qualify for participation in the Company's bonus and incentive programs.
