AI Systems Architect

Cell biology is where computing was in 1975\. The tools that drive drug discovery and regenerative medicine are room-sized machines costing millions of dollars, operated manually by researchers working nights and weekends. Protocols disappear when the postdoc leaves. Data sits in local drives… unstructured, unlabelled, and unshareable.

We are changing that. We’ve built a compact, AI-native hardware platform that automates the full cell culture workflow: stem cell expansion, 3D organoid formation, and drug assay.

We have 8 prototypes running in labs across the Bay Area, 20\+ active users, 5 peer-reviewed publications, and placements in ecosystem.

The science works. The hardware works. Now we’re building the intelligence layer.

The Role

We are hiring an

AI Systems Architect

to design and build the intelligence layer of the platform – the architecture that connects devices (sensors, actuators, fluidic systems, imaging hardware) to NVIDIA GPU compute, cloud services, and AI models. You will turn a biology lab into a living software system.

This is a senior individual contributor role reporting directly to the co-founders.

This is a founding team role with equity compensation reflecting the stage and your impact.

What You’ll Be Doing

This is a startup. The scope is wide on purpose. You will move across these areas based on what matters most, week to week.

Build the Cloud

This is the nervous system of the platform, connecting devices across labs, handling protocol execution and scheduling, routing biological signals to AI processing nodes, and piping structured data into models. You will architect this from the ground up.

• Design and build the

IoT messaging architecture

that connects disparate physical instruments (microscopes, MEA electrophysiology rigs, fluidic pumps, temperature controllers) into a unified, addressable system – building on our existing MQTT/AWS infrastructure and scaling it to production

• Build

NVIDIA GPU compute pipelines

that process raw sensor streams (image sequences, electrophysiology recordings, fluidic telemetry) in real time and route outputs to downstream AI models

• Design the

biological data schema

– structured, versioned, labelled, and shareable – that underpins our long-term vision of a verified cell culture data marketplace

• Connect the hardware control layer to the

AI-native operating system

, so the device is not just automated but agentic: learning from runs, correcting errors mid-experiment, and flagging anomalies before they become failures

Build the Digital Twin

Digital twin of a cell culture: given hardware state, reagent properties, protocol parameters, and sensor data at time T, the model predicts biological state at time T\+N and uses that prediction to adapt the protocol in real time.

• Develop the

feedback control architecture

that enables real-time, closed-loop protocol adaptation based on live sensor data (imaging, electrophysiology, fluidics)

• Build and deploy

on-device inference pipelines

running on embedded NVIDIA hardware (not just cloud-based) for real-time decision making at the bench

• Develop

protocol versioning and reproducibility infrastructure

: every run produces a versioned, reproducible, shareable protocol that any user can download and run

Own the Hardware-Software Interface

• Work closely with the technical founders (CEO \& CTO) to define and maintain the software interface with physical control systems (fluidics, valves, sensors, imaging)
• Architect the

agent framework

(MCP-based) that connects vision models, LLMs, robotics, and cloud services into a coherent, AI-native laboratory operating system

• Contribute to platform decisions that will define the product for years, more than implementing a spec, you are narrating it

Must-Haves

• Strong experience architecting

IoT or distributed sensor systems

, connecting physical hardware to cloud infrastructure at scale

• Hands-on experience with

NVIDIA GPU compute

in a production context (Jetson, A100, H100 or equivalent); comfortable designing inference pipelines for real-time sensor data

• Demonstrated ability to build

AI/ML pipelines end-to-end

: data ingestion, model training or fine-tuning, deployment, monitoring

• Experience with

cloud infrastructure

: AWS (S3, IoT Core, Glacier, or equivalent), containerisation, and orchestration (Docker, Kubernetes, or similar)

• Experience with

MQTT or equivalent pub/sub messaging protocols

for real-time device communication

• Strong software engineering fundamentals – you write clean, testable, documented code and think carefully about system design

Nice-to-Haves

• Experience with

agent frameworks and MCP (Model Context Protocol)

, connecting LLMs and vision models to real-world tools, APIs, sensors, or robotics

• Background in

robotics or mechatronics

, understanding how physical actuation systems interface with software control layers

• Familiarity with

biological or biomedical data

, electrophysiology (MEA), microscopy image analysis, multi-omics, or microfluidics

• Experience with

Nextflow, Dockstore

, or similar workflow management tools

• Exposure to

digital twin methodologies

or model-based systems engineering

• Existing relationships with

CROs

(Charles River, PPD/Thermo Fisher, Labcorp),

pharma

(AstraZeneca, Merck, Genentech), or

academic partners

(Stanford, UCSC, Northwestern)

What We’re Looking For in a Person

Beyond the technical skills, we are hiring for three qualities that define every great early-stage team member:

High autonomy.

You take ownership. You don’t wait to be told – when something needs solving, you solve it and report back. You actively seek more responsibility, not less.

High openness.

You stay current with fast-moving AI and hardware fields as a matter of course, not as a job requirement. You’re adaptable and genuinely excited by the unknown.

High conscientiousness.

You understand the implications of your decisions. You think about the big picture. You can sensibly weigh risk against speed, and you communicate that reasoning clearly to the team.

Why Now?

The FDA Modernization Act 2\.0 has removed the two-animal requirement for IND filings, putting human-relevant in vitro models at the centre of preclinical drug development. The NIH has opened an $87M Standardised Organoid Modeling centre. Charles River called 2025 “the year of alternative methods.” The preclinical CRO market is $8\.62B, largely manual and underautomated.

We are early, well-positioned, and moving fast!!