Member of the Technical Staff - AI/ML Engineer

Transfyr•Cambridge, MA

1d•Onsite

About The Position

AI/ML engineers at Transfyr build the learning systems that turn raw observations of scientific work into usable insight, feedback, and automation. You will build end-to-end ML systems that learn from messy, real-world data captured in active laboratory environments. You’ll work closely with our computer vision team to integrate grounded data (e.g. object coordinates, action timings, etc.) to develop grounded interpretations of actions and their scientific impact. Your models must contend with partial observability, significant noise, long context requirements, changing protocols, and ambiguous outcomes and still produce signals that scientists and downstream systems can trust. The role demands strong ML fundamentals, solid software engineering judgment, and high agency. You will work closely with perception engineers, software engineers, and scientists to ensure models are grounded in reality and tightly integrated into real workflows. We’re tackling frontier-hard AI problems and applying those models to frontier science. We're building a team, and we have needs across levels, from hands-on builders early in their careers to senior engineers who enjoy shaping learning architectures and technical direction. This role is in-person in Cambridge, MA (other locations may open in the future, feel free to reach out even if Boston is not currently an option for you).

Requirements

High agency. You don’t wait for perfect datasets or well-posed problems. You identify what needs to be learned, build the right scaffolding, and push work forward.
Biased toward action. You prototype quickly, test assumptions against real data, and iterate based on failure rather than waiting for theoretical certainty.
Successful in ambiguity. You can make progress when labels are incomplete, feedback is delayed, and success criteria evolve over time.
Thoughtful. You understand when sophistication helps and when it obscures, and you make deliberate tradeoffs between model complexity, robustness, and operational cost.
Clear, direct communicator. You can explain model performance and limitations to collaborators across engineering, science, and operations.
Intense. You care deeply about the mission, work hard when it matters, and help keep the team oriented toward what actually moves the needle.
Strong programming expertise with experience in software engineering, data systems, and AI/ML product development.
Strong grounding in machine learning, with experience building models that learn from noisy, real-world data rather than clean, static datasets.
Experience working with or reasoning about multimodal systems (e.g., vision, audio, sensor data, metadata, text), and an intuition for how different signals complement or confound each other.
Fluency in Python and modern ML frameworks (e.g., PyTorch), with experience training, evaluating, and iterating on models in real systems.
Experience designing data pipelines and training/evaluation infrastructure that evolve over time as new data arrives and assumptions change.
Understanding of what it takes to move from prototype to production, including monitoring, iteration, and maintaining models as environments drift.
Ability to work across the stack in close collaboration with software and perception engineers, understanding that model performance depends on the surrounding system.
Strong fundamentals, curiosity, and the ability to quickly learn new tools, models, or domains as the problem demands.

Nice To Haves

A passion for and experience in science
Entrepreneurial spirit - hackathons, successful projects, startups
Demonstrated experience working in fast-moving/ambiguous environments (like startups!)

Responsibilities

Build ML systems that learn from real-world scientific execution data where feedback is delayed, labels are incomplete, and outcomes are confounded by how work was actually done.
Collaborate with perception engineers to design multimodal learning pipelines that combine vision, audio, sensor data, metadata, and outcomes into coherent representations of scientific workflows.
Develop models that can reason about why an experiment succeeded or failed when intent, execution, environment, and outcome are tightly entangled.
Build systems that surface model confidence / uncertainty, enabling scientists to understand when to trust a recommendation and when to intervene.
Integrate models into real workflows where outputs influence both human decisions and robotic actions, and model behavior must remain robust as protocols, operators, and environments change.
Ensure models learn transferable structure rather than lab- or site-specific artifacts, enabling insights to carry across experiments, teams, and geographies.
Enable future physical AI systems by ensuring models learn from execution-level data, not just outcomes, building foundations for automation that can work in the real world.