Evaluation infrastructure
Harnesses, logging, evidence capture, reproducibility tooling, and structured outputs for adversarial evaluation.
Talent
Syntony Engineering Fellowship
Syntony · Remote · Durham, NC · Formation-stage startupA selective technical fellowship for exceptional engineers, researchers, and builders. Work directly with Syntony's founder on AI evaluation infrastructure, agent systems, risk data, and research tooling. A serious path for people who want more than a conventional internship — and, for exceptional contributors, a possible path into future full-time engineering roles.
About Syntony
Syntony is an AI risk firm. We conduct adversarial evaluations of frontier and production AI systems, build the governance architecture to act on findings, and ship software that makes both repeatable. We work with frontier labs, public sector clients, and enterprise deployments where the stakes are real.
Syntony is currently founder-led and in its formation stage: pre-seed, building the first repeatable systems across AI evaluation, governance tooling, and risk intelligence. That means fellows will see unusually close-range company-building: technical prioritization, product judgment, research direction, and the early shape of the engineering culture.
What this fellowship is
The Syntony Engineering Fellowship is a selective, part-time technical fellowship for people who want to do unusually substantive work at the intersection of engineering, AI evaluation, governance, and applied research.
Fellows work directly with Nathan Heath, Syntony's Founder & CEO, on bounded prototype and research engineering projects. The fellowship is designed for people with strong technical taste, independent judgment, and the desire to build systems that matter before the category is fully mature.
This is not a conventional internship. It is closer to a miniature research lab apprenticeship: high-context, high-trust, and built around real technical artifacts.
Technical taste over credentials
We care about what you can reason through, build, test, and explain. Strong candidates may be students, recent graduates, independent builders, researchers, or engineers outside the usual pipeline.
Substantive but bounded work
Projects are scoped so fellows can own meaningful technical work without becoming unpaid staff. The goal is visible progress, clear learning, and artifacts you can stand behind.
Direct founder collaboration
Fellows work closely with Syntony's founder on technical direction, research questions, product judgment, and engineering execution.
Pathway into future roles
Exceptional contributors may be considered for future full-time engineering roles as Syntony grows. The fellowship does not guarantee employment, but it is intended to be one serious path into the team.
Project areas
Where fellows may build
Agentic systems and tool use
Testing systems that plan, delegate, use tools, retain memory, and act across multi-step environments.
Risk data and index engineering
Data pipelines, scoring systems, taxonomies, and visualizations for emerging AI risk signals.
GovTune AI prototypes
Research and prototype work around turning red-team traces into governance findings, review artifacts, and decision support.
Causal modeling and system dynamics
Tools for mapping feedback loops, escalation pathways, and cross-domain AI risk interactions.
Research engineering
Small, sharp prototypes that help turn ambiguous research questions into testable systems.
Strong candidates will have
- Strong programming ability, especially in Python or TypeScript.
- Evidence of independent building: GitHub projects, research code, products, demos, technical writing, or serious course or research work.
- Clear written communication and the ability to explain technical choices.
- Interest in AI safety, evaluation, governance, red teaming, systems thinking, or strategic risk.
- Good judgment around ambiguity, confidentiality, and scope.
- A desire to work on problems where the answer is not already written down.
Bonus
- Experience with LLM evaluation frameworks, agent scaffolds, data pipelines, or developer tools.
- Prior work in AI safety, ML security, policy, forecasting, or governance.
- Open-source contributions, research publications, demos, or technical essays.
- Comfort reading papers and turning ideas into working prototypes.
- Taste for simple systems that actually get used.
Structure & pathway
How the fellowship works
Part-time and project-scoped
Time commitment, duration, attribution, confidentiality, IP terms, and any stipend or compensation arrangement are agreed in writing before work begins.
Not a full-time role
This is not a full-time employment role and does not guarantee future employment. The fellowship is designed around learning, research, prototypes, and high-signal collaboration.
Possible path into the team
Fellows who demonstrate unusually strong technical judgment, independence, and alignment with Syntony's mission may be considered for paid engineering, founding-team, contractor, or long-term collaborator roles.
Separate scope for client work
Client-facing work is not assumed. If a project becomes substantial commercial product work or client delivery, that work will be handled through a separate written agreement.
How we evaluate
We weight demonstrated work over credentials. Lead with your strongest work: code, demos, papers, technical writing, products, or public analysis.
- Application review.
- Short async conversation.
- Technical conversation with Nathan.
- Small scoped project discussion or work sample.
- Fellowship invitation and written project scope.
Apply
Apply to the Syntony Engineering Fellowship
Applications are reviewed on a rolling basis. We read these carefully. Lead with your strongest technical work: code, demos, papers, products, technical writing, or public analysis.
We are especially interested in what you have built, how you think, and what kind of work you would want to own at Syntony.