Careers
Build the layer that lets AI act.
Intelligence is becoming a commodity; the durable product is the runtime governance that makes an agent trustworthy enough to act — every action checked in the path before it executes. We build that, and the agents that run on it.
The team we're building toward
We're early, and we're not hiring these roles yet — but we know the shape of the team that runs and builds runtime governance, because it's the same operating model we build for our customers.
When AI agents do the work, the human roles move up the stack: someone sets the bounds, someone handles the exceptions, someone proves it all behaved. These are the roles we'll grow into — and the kinds of people we'll want when we do.
None of these are open today. If one is clearly you, introduce yourself — we'll start here when we open them.
Engineering
Build the runtime — and the agents that run on it.
The product is the gate: every agent action checked against its capability profile, policy, and live session state in the path before it executes — fail-closed, signed, replayable. You'll build that runtime and the agents on top of it.
Principal Runtime EngineerEngineering · Runtime
Own the in-path evaluation engine — the hot path that checks every agent action against its capability profile, policy-as-code, and live session state, fail-closed, in single-digit milliseconds. This is the core of the product.
What you'll do
- Design and own the request-path gate: capability + policy + state checks evaluated inline before any action executes, under a hard latency budget (~8 ms).
- Build the cross-agent session graph so the gate reasons about multi-step and multi-agent behavior, not just isolated calls.
- Make fail-closed real — define correct behavior under timeout, partial state, and overload, and keep it correct at scale.
- Drive the performance work (profiling, caching, concurrency) that keeps governance invisible to the agents it governs.
What you'll bring
- Deep experience building low-latency, high-reliability backend systems (Rust / Go / similar) where correctness under failure matters.
- You've owned a production hot path and have a real bias for fail-closed, fail-safe design.
- Strong distributed-systems instincts: concurrency, consistency, back-pressure.
- Pragmatism about policy and security — you don't need to be a compliance expert, but you respect the constraint.
Not open yet. If this is clearly you, introduce yourself ›
Senior Governance EngineerEngineering · Policy & detection
Turn governance into code. You'll build the policy engine, the capability-profile model, and the detection logic — confidence-laundering, interpretation-divergence, multi-turn crescendo attacks across the session graph — that the runtime enforces.
What you'll do
- Design the policy-as-code model and capability profiles — what an agent is allowed to be, not just which tools it holds.
- Build action-class and effect annotations plus reversibility-gated HITL — the logic that decides what auto-executes versus what a human approves.
- Implement aggregate detection across the session graph: sub-threshold splitting, multi-turn escalation, interpretation divergence between agents.
- Produce the signed, replayable attestation trail that compliance and incident review depend on.
What you'll bring
- Strong backend engineering plus genuine interest in the semantics of policy, authorization, and trust.
- Experience with rules/policy engines, authz systems, or static/dynamic analysis is a plus.
- You can turn a fuzzy rule into a deterministic, testable specification.
- A security mindset — you think about how a system is abused, not only how it's used.
Not open yet. If this is clearly you, introduce yourself ›
AI Agent EngineerEngineering · Agents
Build the agents that run on the governed runtime — our horizontal CX agent and the insurance agents (Atlas, Praxis). You'll ship agents that do real, consequential work and are governed by AO from the first action.
What you'll do
- Design and ship production agents on foundation models: tool use, multi-step planning, escalation, and structured outputs.
- Integrate agents with the AO gate — capability profiles, effect annotations, the oversight inbox — so every action is checked in the path.
- Connect agents to real systems of record (claims platforms, CRMs) with safe, reversible writes.
- Tune for faithful escalation: the agent acts when it's sure, asks when it isn't, and never acts wrong silently.
What you'll bring
- Hands-on experience building LLM agents or tool-use systems that reached real users.
- Strong applied engineering (TypeScript / Python) and a feel for prompt, eval, and guardrail design.
- Product instinct for where a human should stay in the loop.
- Curiosity about the domains we serve — insurance first.
Not open yet. If this is clearly you, introduce yourself ›
Operations & oversight
Run governed agents in production.
We don't just sell the runtime — we operate governed agents for customers. These are the new operating-model roles: when the agent does the work, the human moves up the stack — from doing the task to setting the bounds and owning the exceptions.
Agent Operations LeadOperations · Agent operations
Own the day-to-day operation of governed agents in production across customer deployments. You set the thresholds, watch the fleet, and decide what graduates from human-approved to automatic — the human side of “AI does the work.” (This is agent operations, not ML infra / MLOps.)
What you'll do
- Run agent operations across deployments: monitor outcomes, escalation rates, and where agents ask versus act.
- Tune capability profiles and confidence thresholds with the governance team; decide what auto-graduates to a rule and what stays gated.
- Own the oversight-inbox workflow — staffing, SLAs, and the feedback loop from human decisions back into policy.
- Be the operational voice in customer reviews: what the agents did, what they escalated, what changed.
What you'll bring
- Experience running an operations, support, or trust-and-safety function at scale.
- Comfort with data — you reason about rates, thresholds, and trade-offs, not vibes.
- Calm judgment under ambiguity; you decide where the line between automatic and human sits.
- Clear distinction in your head between overseeing acting agents and running ML infrastructure.
Not open yet. If this is clearly you, introduce yourself ›
AI Oversight OperatorOperations · Human-in-the-loop
The front line of human-in-the-loop. You work the oversight inbox: review the actions agents escalate, approve, hold, or deny them, and leave the reasoning that sharpens the policy. The agent handles the volume; you own the judgment on what matters.
What you'll do
- Review flagged agent actions in the oversight inbox and make the call — approve, hold, redirect, or deny — with the context the gate surfaces.
- Handle the high-stakes and ambiguous cases agents are designed to escalate rather than guess.
- Document the why behind each decision so it becomes policy and reduces future escalations.
- Surface patterns — recurring escalations, edge cases, drift — to the operations and governance teams.
What you'll bring
- Sharp judgment and strong written reasoning; you can justify a decision clearly and quickly.
- Domain experience in a regulated or operational field (insurance, healthcare, finance, support) is a strong plus.
- Comfort working a queue with SLAs and owning the outcome of each call.
- Integrity — you are the human accountability in the loop.
Not open yet. If this is clearly you, introduce yourself ›
Risk & assurance
Make “trustworthy enough to act” provable.
Trust is the product. This family owns the evidence — that the runtime governance maps to the frameworks our customers answer to, and that every action can be proven after the fact.
AI Compliance OfficerRisk & assurance
Own how Bitligence's runtime governance maps to the frameworks our customers are measured against — SOC 2, ISO 27001, HIPAA, the EU AI Act — and turn the signed, replayable trail into audit-ready evidence. You make the product's trust claims provable.
What you'll do
- Map the runtime's policy model and attestation trail to control frameworks (SOC 2, ISO 27001, HIPAA, EU AI Act) and keep the mapping current as regulation moves.
- Build the audit-ready evidence customers need from the signed ledger — what to retain, how to present it, how to answer an auditor.
- Partner with governance engineering so the right artifacts are produced in the path, by design.
- Advise regulated customers on how to evidence their AI controls using AO — we help customers meet standards; this role owns that, and our own posture.
What you'll bring
- Experience in compliance, audit, GRC, or model risk — ideally touching AI/ML or software controls.
- Working knowledge of one or more of SOC 2 / ISO 27001 / HIPAA / the EU AI Act.
- Ability to translate fluently between regulators, engineers, and customers.
- Precision and honesty about what is — and isn't — being claimed.
Not open yet. If this is clearly you, introduce yourself ›