Help us buildTOTAL.
See open roles ↓Our culture is intense and our pace is relentless. We work in short sprints with deadlines every week, driven by customer requests.
Our CEO talks directly with engineers all day, every day. Our product owners spend most of their time writing code, not specs. We have no room for high-level product people who want to steer without building.
We’d rather build it than talk about it. The fastest way to win an argument here is to make it real. Every successful hire we’ve had shipped a feature in their first week.
Who we need
Persona AI Engineer
You own the Generative Persona, the behavioral foundation model TOTAL trains for every identity. It learns how each identity behaves and forecasts the next action before it happens. Every verdict TOTAL renders stands on what you build.
What you’ll do
- Train the behavioral foundation models that forecast each identity’s next move
- Push the architecture forward with new signals and sharper forecasts
- Build the training and eval loops that catch drift before customers do
- Bring rigorous experimentation: clean ablations, honest baselines, no superstition
- Own a slice of inference end to end, from offline training to real-time predictions on production traffic
What we’re looking for
- You train models because you want to, not because a roadmap told you to. Show us something you built and launched
- Real depth in modern ML, from raw data to a trained, served model without hand-holding
- Speed. Paper to prototype in a weekend, prototype to production the moment it earns it
- You live in the modern AI stack. RAG, agentic workflows, fine-tuning, modern serving and evals. New models and tools land in your work the week they ship
AI Security Engineer
You build the Judge, the agentic workflow that reasons like a court to turn a predicted intent into a defensible verdict. It gathers evidence from the live environment, grounds it against an attack-vector corpus, and reasons until it can defend the call: halt or let through.
What you’ll do
- Build the Judge’s agentic workflow: gather evidence, ground it against the attack-vector corpus, synthesize one verdict with citations
- Design and stand up the evaluation harness: red-team prompts, latency budgets, regression suites
- Improve the Judge’s calibration with measurable reductions in false positives and false negatives, every release
- Work directly with security analysts to test the system against real adversarial patterns
- Route each case to the expert built for its threat family, and give the hard calls more room to reason
What we’re looking for
- You build with LLMs for real. RAG, agents, tool-calling, fine-tuning. Show us an agent you put in production, not a notebook
- Fluent in the modern stack: vector stores, retrieval eval, and harnesses you actually trust
- You think like an attacker, or you’re hungry enough to get there fast. Threat models, attack taxonomies, MITRE ATT&CK
- You get features in front of real users and measure them. Fewer false positives, lower latency, every release
DevOps Engineer
Keep the platform fast and always on. You own the infrastructure that runs TOTAL’s agentic fleets in-network, at machine speed, so the verdict lands in real time across Fortune 500 deployments.
What you’ll do
- Own the production stack end to end: Kubernetes, observability, secrets, CI/CD, deploys
- Cut latency in the inference path. Every millisecond matters when the verdict is real-time
- Harden the platform with zero-downtime deploys, blast-radius isolation, and fast rollbacks
- Stand up enterprise customers in days, not weeks, on infrastructure that never has to reach out
- Run TOTAL’s agentic fleets at scale: orchestrate the subagents behind discovery, clustering, and the Judge, all in-network
What we’re looking for
- You’ve run real systems under real load. Kubernetes, Terraform, AWS or GCP, and the scars to prove it
- Latency is personal. You chase milliseconds in the inference path because real-time means real-time
- Observability fluency. Prometheus, Grafana, OpenTelemetry, whatever it takes to see inside a live system
- You move fast without breaking prod. Zero-downtime deploys, fast rollbacks, blast-radius thinking
Full Stack Engineer
Build end to end, from the data models that capture identity behavior to the console where analysts weigh the Judge’s verdict and act. The features you ship turn raw signal into a decision, and the adjudications they capture feed the loop that sharpens the system.
What you’ll do
- Own features end to end: schema, API, UI, deploy
- Work in the inference path. Wire new signals into the Persona model and surface their effects in the console
- Build the integrations that connect TOTAL to the identity providers, EDR systems, and SIEMs already in the stack
- Own quality from prototype to GA. Your features go to customers under your name
- Build the console where an analyst weighs the Judge’s verdict and acts on it, from a self-police nudge to lockdown
What we’re looking for
- You move end to end and fast. Schema to API to UI to deploy, all yours
- Fluent across the stack. TypeScript, Python, or Go, with a real database underneath
- Modern web in your bones. React, clean APIs, a feel for when to add a dependency and when not to
- You’ve made things people use. Side projects, startups, whatever proves you build
Behavioral Research Scientist
A Persona is a behavioral model of a person, and it’s only as sharp as the science under it. You own the psychology behind both halves of the system: what a Persona captures about how someone behaves, and how TOTAL intervenes when behavior drifts, from a self-police nudge to the moment it has to escalate. The science you define is what every model and every intervention downstream is built on.
What you’ll do
- Define the behavioral and psychological principles a Persona is built on, and defend why each one earns its place
- Turn research into something the models can learn from, from hypothesis to signal to validated dimension
- Validate the science against real enterprise behavior at scale, not lab conditions
- Work shoulder to shoulder with the Persona engineers to close the loop from theory to production
- Design the science behind the intervention: what nudge pulls a drifting identity back, and when a self-police prompt stops being enough
What we’re looking for
- Deep grounding in psychology or behavioral science. You reason about mechanisms, not vibes
- You’ve turned research into something real: a model, a measure, a finding people actually use
- Rigor. Clean study design, honest baselines, results you can defend
- You want your research to move a product, not sit in a journal
Don't see your role?
If you read this and already know what you'd build, tell us. We make room for exceptional talent.
Contact jobs@truu.ai