Senior Staff AI Engineer

The Opportunity

We are looking for a Senior Staff AI Engineer with 8-15 years of experience to own the technical direction of our agentic AI harness and transform how chemists and AI reason about science together. You will be the most senior hands-on IC on the Agentic AI team, building the orchestration, tool use, memory, and reasoning layer at the center of Albert OS — the first AI-native operating system for materials science. This is a 0-to-1 role where you'll define how the entire platform shifts from a chat-based interface to an agentic-first experience, working with trillions of proprietary scientific tokens that no one else in the world has access to. You'll lead two junior engineers on the AI harness while partnering closely with product and ML leadership to expand agent capabilities across every surface of the platform.

What You'll Be Doing

• Build and expand the agentic harness (think Claude Code or Devin, but for chemistry) — designing orchestration, tool use, memory, planning, and evaluation layers for production agent systems
• Set the technical direction for shifting the entire Albert OS platform from a chat-style interface to an agentic-first experience where agents operate across every surface
• Develop agentic tools and sub-agent personas that operationalize how chemists think and work in their day-to-day workflows
• Integrate frontier reasoning models into the application, expanding their capabilities and potentially contributing to fine-tuning programs for domain-specific reasoning
• Make agent behavior observable and measurable through tracing and evaluation against real scientific tasks, driving reliability to the level where partners trust agents to act autonomously

About the Company

Meet Albert, the partner for the scientist of the future.Enterprises partner with Albert to reimagine how they invent. From digitalizing R&D and managing change at scale, to unlocking entirely new business models with AI, chemical and materials science leaders rely on Albert to achieve digital transformation — and ensure it delivers lasting value. Every day, scientists in 30+ countries use Albert to accelerate R&D with AI trained like a chemist, bringing better products to market, faster. Your science moves the world forward. We move everything else out of your way.

Founded 2021

Website albertinvent.com

Total funding $45M

Office locations Oakland, CA

Company locations San Francisco Bay Area

Founded in 2021.

About the Team

The AI and ML team at the company is an 11-person organization within a ~200-person company, sitting at the core of everything that makes Albert's chemistry-native AI platform possible. The team spans two focus areas — Agentic AI and Machine Learning — each led by a dedicated director, with the broader AI org reporting up through the VP of AI & ML.

The Agentic AI team is building what is essentially Claude Code or Devin, but for chemistry: a production harness that orchestrates agent workflows, proposes hypotheses, suggests next experiments, and enables chemists and AI to reason about science together. The team includes AI engineers working on the harness itself alongside full-stack engineers building the application layer of the agent system.

What makes this team unique is the blend of deep AI engineering talent and scientific sensibility. Several team members hold advanced degrees in physical sciences, and the culture prizes engineers who can put themselves in the shoes of a working scientist — not just write great code, but understand why the system they're building matters to someone at the bench. The team also has a program to fine-tune reasoning models on proprietary scientific data, pushing the frontier of what AI can do in materials science.

The team operates in a fast-paced, 0-to-1 environment where engineers are expected to own sophisticated systems from inception through production. The current focus is on expanding agents beyond the chat interface to operate across every surface of Albert OS — transforming the entire platform into an agentic-first experience. It's a high-impact, high-autonomy environment where the work directly shapes how the world's largest chemical companies invent.

The Product

Albert OS is the first AI-native operating system for materials science — a chemistry-native platform that transforms how the world's largest chemical and materials companies invent.

The platform works end-to-end across the R&D lifecycle. Albert goes into analog laboratories still running on paper notebooks, sticky notes, and index cards, and transforms them into fully digitized environments with iPads, computers, and projectors — eliminating paper entirely. This is the foundation that turns proprietary experimental data into each organization's primary strategic asset.

On top of this digitized data layer, Albert deploys AI trained like a chemist — agentic systems that can orchestrate workflows, propose hypotheses, suggest next experiments, and reason over a three-layer scientific data model. The platform operates as a collaborative surface between humans and AI: agents suggest, scientists verify, and experiments get dispatched into the lab. Today this runs through an advanced agentic chat interface, and the product is rapidly expanding so that agents operate across every surface of the platform — proactively acting on data, carrying goals across sessions, and serving as each partner organization's own specialized scientific intelligence.

What makes Albert different: Most AI-for-chemistry companies focus on accelerated molecular discovery — roughly 3–5% of the chemical industry. Albert targets the other 95%: formulation engineering, the chemistry behind everything from paint and glass coatings to personal care products and adhesives. This data lives inside enterprises as proprietary IP, invisible to frontier models trained on public data. Albert's deep embedding inside these organizations gives it access to trillions of proprietary tokens that no other company — and no foundation model — will ever see.

Scientists in 30+ countries use Albert every day to accelerate R&D and bring better products to market faster.

Tech Stack

Python, LangGraph, LangChain, Agentic Frameworks, Frontier LLMs (GPT-4, Claude, etc.), Braintrust, Tracing & Evaluation Systems, RAG, React (nice to have), TypeScript (nice to have)

---

Application Questions

1. Are you able to work in Oakland and come into the office 3 times per week?
2. Walk us through a production agent harness you've built from inception through to production
3. What's your background in physical science (physics, chemistry, biology, materials engineering, etc.) and how has it shaped the way you test and evaluate the AI systems you build?