Careers at Inferal

We're Hiring

Help businesses stop turning decisions into software projects.

Software was never the goal.

Every business runs on decisions. When a shipment runs late, tell the customer. When a trade breaches a limit, stop it. Simple to say. But to make it happen, the decision becomes software.

The moment a business requirement becomes software, it inherits every software-shaped problem. Bugs, dependencies, security patches, upgrades, migrations: a whole lifecycle the original decision never had. By the time it runs, the reasoning behind it has mostly fallen away. The lifecycle stays.

AI makes writing software cheap, but writing was never the expensive part. Every new piece of software is one more thing to operate, secure, upgrade, and eventually retire. You can generate code faster; you can't make its lifecycle disappear. Scaling the creation of software just scales the cost of keeping it alive.

The goal was always the decision and the outcome it was supposed to produce. Your systems already know what happened. The missing piece is something that knows what those facts mean across systems, what follows from them, and why, without turning every decision into another software project.

What we're building

Inferal keeps operational policy connected to the live facts it governs. Three products, one loop:

Relay keeps data moving between the systems a business depends on: APIs, databases, and event streams, with rate limits, credentials, retries, ordering, and history handled. It runs in production today, and paying customers build their agents on it.

Operational Models make meaning explicit: entities, relationships, policies, and constraints, built on RDF and small, composable ontologies we publish through Ontosome.

The Data Platform evaluates what changing facts imply and delivers the resulting actions, alerts, or context to the people, systems, and agents that can act. Every outcome stays connected to the evidence and policy behind it.

The hard problems

These are real, open problems. We have approaches, not answers. If that distinction matters to you, keep reading.

  1. Integrations that build themselves

    Every provider has its own rate limits, auth quirks, pagination, and broken webhooks. We generate integrations from evidence such as API specs and documentation, then validate them until they behave. The goal: onboard a new source without a person in the loop, and keep it fresh at scale.

  2. A data-native implication engine

    Evaluating rules and constraints continuously over live, RDF-shaped operational models. Incremental, so one changed fact doesn't recompute the world. Time-aware, because deadlines and durations are business logic too. And every conclusion carries its provenance.

  3. From model to running system

    Turning a declarative operational model into a physical execution plan: scheduling, placement, and delivery across heterogeneous runtimes, from Cloudflare Workers to a customer's own Kubernetes cluster.

  4. A million ontologies

    Instead of one universal model, a large library of small, scoped ontologies and mappings that can be created, tested, composed, and maintained like software, with and without LLM assistance.

  5. Interaction, not interface

    When decisions stop living inside applications, people and agents need new ways to see, question, and steer them. We're working out what that looks like, starting with the person, their role, and their goal instead of screens.

Who we're looking for

Nobody here has just one job. We don't divide the work into "I do backend" and "you do design." We work in projects, and each project needs whatever it needs: systems work, modeling, interfaces, customer conversations. You bring real depth in something, and you pick up the rest along the way. Your background is where you start, not where you stop.

The people we're looking for tend to arrive from one of these directions, and to be curious about the other two:

Close to the metal
Engineers who have built storage, query or rule engines, schedulers, or integration infrastructure, and know that a system customers depend on at 3 a.m. is different from a demo. Rust, TypeScript, Python, and whatever else the problem demands.
Meaning
Ontologists and knowledge engineers who can make meaning explicit enough for software to act on: scoped ontologies, identity criteria, constraints, provenance, and mappings to messy source systems. Experience with RDF, OWL, or SHACL helps, but judgment matters more than allegiance to a stack.
Human-computer interaction
People who think about how people and agents actually work together: how a decision gets seen, questioned, and trusted, and what replaces the application when the application is no longer the center.

We care more about what you've built than where you went to school. More about how you think through a problem than which titles are on your resume. Attitude first, then belief, then skill. Skills can be learned. Attitude can't.

Not ready to apply yet? Join the Software Internals Discord, a community where people dig into the kinds of problems listed above.

Open roles

Each role starts from one of the directions above and grows into the others. The cards describe where you'd begin, not the edges of the job.

Software Engineer

EngagementFull-time or contractLocationSan Francisco or remote

You'll build the systems that keep operational meaning connected to live data: integrations that maintain themselves, an incremental implication engine, and the runtime that runs operational models anywhere from Cloudflare Workers to a customer's Kubernetes cluster.

We need depth and breadth. You've gone deep on something real, such as a storage engine, a query planner, a scheduler, or a compiler, and you move across the stack when a project needs it. One week that means profiling a hot path or reading a protocol spec; the next, shaping an API or an ontology.

What you'll do

  • Design and build core infrastructure for Relay, the Data Platform, and Operational Models
  • Go low-level where it matters: memory, concurrency, I/O, and performance under real load
  • Model domains with ontologies and turn them into systems that evaluate and act
  • Work directly with customers on the problems your code solves

What you bring

  • Deep experience in at least one area of systems work, and the range to work across the stack
  • Familiarity with ontologies and RDF is required: triples and named graphs, SPARQL, and OWL or SHACL
  • Rust, TypeScript, or Python, and the judgment to choose what the problem needs
  • Systems you've shipped and kept running for people who depended on them

Applied ML Engineer

EngagementFull-time or contractLocationSan Francisco or remote

You'll build the data and model side of Inferal: datasets drawn from operational models, ontologies, and integration evidence, and the models trained on them to generate integrations, map source systems to shared meaning, and propose rules people can inspect.

What you'll do

  • Build dataset pipelines: collection, cleaning, labeling, versioning, and provenance
  • Train and fine-tune models, from quick experiments to production runs
  • Design evaluations that tell us whether a model actually got better
  • Ship models into the product and keep them healthy once they're there

What you bring

  • End-to-end experience taking a model from raw data to a deployed system
  • Strong Python and the data and training tooling around it
  • Rigor about evaluation, reproducibility, and where data came from
  • Experience with RDF, ontologies, or structured data is a plus

Business Development Intern

EngagementPaid internship, hourlyLocationSan Francisco Bay AreaPayHourly, plus transit assistance and a ride-share allowance

Inferal grows through relationships. The teams we work with usually find us through someone they trust, a conversation at an event, or an introduction that arrived at the right moment. You'll build those connections across the Bay Area, nationwide, and globally, and turn them into conversations that matter for the business.

This role shapes who we meet. The people you reach, the rooms you get us into, and the follow-ups you keep alive decide which customers, partners, and future teammates hear about us first.

What you'll do

  • Research companies, people, and communities through email, social media, and the web
  • Write outreach that reads like it came from a person (because it did), and follow up until it becomes a conversation
  • Organize and attend Bay Area events, from meetups and dinners to conferences
  • Travel to occasional events outside the region
  • Keep track of every conversation so relationships don't fall through the cracks

What you bring

  • You embrace AI in your workflow for research, writing, and staying organized
  • Clear writing and comfort starting conversations with people you don't know yet
  • Reliability with details, schedules, and follow-through
  • A software engineering or computer science background is a significant advantage

What to expect

How we work is part of what we build. We keep finding better ways for people and agents to work together, then turn what works into shared tools and practice. Our work spans repositories, SaaS products, and systems we build ourselves. We use each where it works best, and you'll help shape that system.

We prioritize sleep. Great work comes from rested minds, not from heroic hours.

San Francisco office, remote-friendly. Some of us work from our office in San Francisco; some work from wherever they do their best thinking. Either way, we make time to be in the same room with coworkers, customers, and peers.

Flexible engagement. We hire full-time and, where it fits, on contract. Full-time roles include a competitive salary and early options with extended exercise windows for tenured team members.

Further reading

If you want to understand the problem space and how we think about it, these are worth your time:

Apply by video

Join us

We ask for a video because it tells us something a resume can't. Resumes are easy to send to a hundred companies at once. A video takes thought. That's the point: we want to hear from people who read this page and felt something click, not people running a pipeline.

Talk about something you built that you're proud of. Tell us why this problem space interests you. No need for script or polish!

We'll ask for camera and microphone access.