לוח משרות דרושים, לחיפוש זה נמצאו 34,376 משרות

our company secures AI apps, agents and models from build to production. Enterprises are shipping AI faster than they can secure it, attack techniques evolve constantly, and every company draws the line on acceptable behavior in a different place. No enterprise can keep up with that on its own, and our company is how they do it. At the heart of the platform sit two engines: the attack engine finds the failures before attackers do, and the defense engine blocks them in production without breaking legitimate traffic.
We are hiring a Senior PM who will lead the strategy, vision, and execution of our core engines. This PM sits inside the DS workflow - and can read the notebooks, argue about the eval set, query the data themselves - and translate between model reality and customer reality in both directions.
This is not a PM role that writes a PRD and waits. It is a PM role where I pulled the data, heres what moved the F1, and heres what it means for customers is the expected level of answer.
What you'll own
Attack engine roadmap. Coverage and depth of adversarial techniques, novel-attack generation quality, agentic/multi-turn testing, zero-day and model-risk research productization. You decide which harm categories, modalities, and attack families we invest in next, and you defend that call with evidence.
Defense engine roadmap. Classifier quality across policies and languages, the precision/recall/latency triangle, customer-specific tuning, and the path from "a red team finding" to "a guardrail that blocks it." Closing that attack→defense loop is the single highest-leverage thing this role can do.
Evaluation as a product surface. Own how we define, measure, and publish model quality - benchmark datasets, eval harnesses, golden sets, regression gates, and the metrics customers actually see. When a customer asks "how accurate is this," the answer and the evidence behind it are yours.
Data strategy for the engines. What data we collect, label, retain, and feed back into training; how customer feedback (overrides, false-positive reports, session feedback) becomes a training signal; and how all of it stays within our customer data commitments
The DS↔Engineering seam. Model artifacts, schemas, and pipelines. You keep that contract explicit and get research into production.
Day to day
Run discovery with AI Product Owners, security engineers, and CISOs on what attacks they fear and what false positives they can't tolerate
Turn fuzzy customer harm into a testable engine requirement with a measurable bar
Prioritize the DS backlog with the DS lead; make the hard calls on coverage vs. depth vs. latency
Own the quality narrative for sales, solution engineering, and security reviews - backed by numbers you can defend
Partner with design and platform PMs so engine capability becomes a usable product, not a config flag
Track competitive and academic movement in adversarial ML and guardrails, and translate it into roadmap pressure.

We are looking for a Head of RL Gym to own the end-to-end delivery, quality, and scaling of RL Gym programs for our strategic AI partners.
You will work closely with frontier-model teams, Delivery, Product, Sales, Research, and Operations to turn complex customer and model-development needs into high-quality, scalable programs.
What Youll Own
Own partner delivery from scoping through launch, execution, scaling, and iteration.
Translate research and model needs into clear program requirements, workflows, and success metrics.
Work closely with frontier-model teams to identify model gaps and design RL Gym programs that address them.
Own quality standards and quality assurance, ensuring consistent, reliable outputs across programs.
Build the processes, systems, and operating model required to scale programs efficiently without compromising quality.
Partner with Sales on customer discovery, solution design, scoping, and expansion opportunities.
Work closely with Product to translate partner needs and operational learnings into product requirements and roadmap priorities.
Build strong feedback loops across customers, model teams, Sales, Product, and Operations.
Own operational performance across quality, capacity, timelines, cost, throughput, and partner satisfaction.
Build and lead the program management and operations function as the business scales.

we are looking for a Lead Endpoint Researcher.
What You'll Do
Research AI coding agent runtimes, reverse-engineer how tools such as Claude Code, Cursor, Copilot CLI, Codex, and emerging agents execute commands, access files, spawn processes, and interact with the operating system.
Design agent sandboxing architectures that safely constrain filesystem access, networking, process execution, credentials, and other sensitive endpoint capabilities.
Explore and implement OS-native isolation mechanisms across macOS, Windows, and Linux, including containers, namespaces, sandbox profiles, virtualization, restricted processes, and permission models.
Build endpoint "harnesses" that allow TokenShift to intercept, control, observe, and modify agent execution flows.
Design infrastructure for deploying and running local LLMs across heterogeneous endpoint fleets, accounting for operating system, CPU architecture, GPU availability, memory constraints, and hardware acceleration.
Evaluate local inference runtimes and model formats such as llama.cpp, MLX, ONNX Runtime, Ollama, vLLM-compatible runtimes, and emerging alternatives.
Develop mechanisms for model discovery, installation, versioning, updates, health monitoring, and lifecycle management across large numbers of developer machines.
Research hardware-aware model selection and routing, deciding which models can run effectively on Apple Silicon, NVIDIA GPUs, Windows workstations, Linux machines, CPU-only environments, and other endpoint configurations.
Build benchmarks that measure model latency, throughput, memory consumption, startup time, resource contention, and developer experience on real endpoint hardware.
Investigate security boundaries between local models, cloud-hosted models, agents, MCP servers, shells, and enterprise resources.
Collaborate with the core CLI and platform teams to take endpoint research from prototype through production.
Track the rapidly evolving AI agent, sandboxing, and local inference ecosystems and proactively identify technologies should support.

As an Adversarial AI Evaluation Scientist, you will review and validate adversarial agent tasks: indirect prompt-injection (IPI), data-exfiltration, tool-misuse, and related attack scenarios to ensure each one is realistic, fair, deterministic, reproducible, and correctly graded. You will decide whether a scenario genuinely demonstrates a model vulnerability or merely an artifact of how the task was built, and you will combine adversarial-security judgment with statistical analysis and LLM-assisted tooling to find quality issues, analyze attack-success patterns, and recommend or automate improvements.
Responsibilities
Review adversarial tasks for realism and threat-model soundness: is the attackers capability plausible, can the attacker actually reach the poisoned content, and is the target system a realistic one?
Assess attack fairness and attribution: confirm the harm is driven by the adversary (not pre-baked into the system prompt, vouched for by a trusted layer, or otherwise rigged), so a careful agent could plausibly pass.
Validate grading correctness: verify that success/failure criteria, reward logic, and deterministic checks detect the intended harm, with no reward hacks, token-collapse, or evadable gates.
Measure and interpret attack success rate (ASR@K) and its variance; verify deterministic, reproducible behavior across repeated executions.
Judge attack quality - novelty vs. known patterns, sophistication, stealth, and coverage across harm types and target surfaces.
Identify anomalies and systemic issues across large sets of tasks (e.g. scenarios that succeed ~100% of the time against every model - a red flag for an unfair task, not a win).
Define validation guidelines, acceptance criteria, quality thresholds, and release gates for adversarial tasks.
Use LLMs and AI models to analyze attack failures, identify root causes, and assist with task or grader fixes.
Build automated validation, regression, and quality-monitoring pipelines for adversarial task sets.
Work closely with red-teamers, researchers, and task authors to improve attack design, fairness, and reliability.

We are seeking a Senior AI Researcher to lead post-training evaluation, red-teaming, and reinforcement learning (RL) gym audits on open-weight models. The ideal candidate will establish rigorous benchmarking methodologies, evaluate large language models (LLMs) against complex threats like Indirect Prompt Injections (IPI), and construct post-training evaluation pipelines that accurately measure realistic frontier-level security capabilities.
Key Responsibilities
RL Post-Training & Benchmarking
Execute post-training runs (e.g. GRPO) using mainstream open-weight generalist models against security-focused RL environments, targeting threat vectors like Indirect Prompt Injection (IPI).
Reward Diagnostics & Trace Analysis - Analyze live loss curves and rollout traces to identify reward hacking, lazy policy convergence, and flawed or over/under-specified verifiers.
Task & Environment Auditing
Review tasks and multi-turn environments (including tool use, web navigation, and computer use) for realism, threat model accuracy, data distribution, and dataset balance.
Performance Reporting (Gym Cards)
Generate comprehensive evaluation cards detailing hill-climbing performance uplift across checkpoints, failure modes, tokens/turns per rollout, and task-level success rates.
Integration & Orchestration
Integrate dockerized environments (e.g., Harbor format) into internal training frameworks, optimizing reset/statefulness semantics, concurrency, and throughput ceilings.

we are seeking an experienced Malware Research Team Leader to lead a team of malware researchers, combining hands-on research with people leadership. This role presents an opportunity to lead an established team, strengthening its skills through ongoing training and guidance while ensuring high-quality deliverables.
The position balances managerial responsibilities with hands-on technical work (approximately 50% each). The ideal candidate demonstrates a positive, proactive attitude and excels as a reliable team player.
Key Responsibilities:
Mentor and manage a team of malware researchers, including recruitment
Conduct short, focused malware research under tight timelines
Detect and document emerging malware trends
Evaluate and enhance team accuracy and performance
Regularly share knowledge with team members and collaborate across teams and departments
Continuously improve malware analysis and review processes
Develop automation processes for malware identification at scale.

We're looking for an exceptional, hands-on technical leader to join our core engineering team as a true force multiplier and technological anchor for the company. You'll have the flexibility to drive high-impact initiatives from conception to production, own complex technical and AI domains end-to-end, or guide and level up a focused delivery team, operating with a high degree of autonomy in an ambiguous, fast-moving environment.
What You'll Do:
Drive strategic initiatives: lead complex technical projects, AI feature developments, and core architectural components with an ownership mindset.
Architect for AI & data at scale: design robust, scalable infrastructure and pipelines capable of handling massive volumes of cloud infrastructure data.
Bridge product, AI & tech: collaborate closely with Product Managers and data specialists to translate business roadmaps into high-performance technical systems.
Promote engineering excellence: establish development guidelines, lead design and architecture reviews, and raise the bar for code quality across teams.
Mentor & guide: act as a technical mentor to fellow engineers through hands-on collaboration, deep code reviews, and architectural consultation.

our company's RL Gyms team builds adversarial environments and evaluations that the world's leading AI labs use to test and train their most capable agents. Every engagement is a delivery project: a lab specifies what it needs to measure, and we produce the environments, adversarial tasks, and verifiers that measure it, against a deadline and a quality bar the customer sets.
Our customers are the frontier labs. You'll work alongside the people building the models everyone else is reading about, on the problems they haven't solved yet, and what you ship becomes part of how they decide whether an agent is safe to release. Very few roles put you that close to the frontier.
We're looking for a Technical Project Manager to own those projects. You'll run several lab engagements at once, each with its own researchers, engineers, and reviewers. You decide who works on what, you hold the schedule, and you decide whether a batch is good enough to ship.
The field is cutting edge and the projects change shape faster than any playbook can describe.
What you'll do
Project ownership
Own concurrent lab engagements end-to-end: scope, milestones, delivery, and status
Turn a customer specification into a task plan with your researchers and engineers: how many tasks, in which environments, against which models, by when
Track every batch through the pipeline (in progress, under review, approved, delivered) and keep the delivery record accurate enough that anyone can read the state of a project in a minute
Surface a slipping deadline early, with the tradeoff already worked out: cut scope, add capacity, or move the date
Researcher allocation and quality
Allocate researchers, task writers, and reviewers across projects based on throughput and specialty
Own their onboarding: brief them on the environment, the attack pattern, and the acceptance criteria, then get them to their first accepted task fast
Give quality feedback on rejected work and spot the patterns behind repeat rejections
Manage the review chain so that no batch reaches the customer before it has cleared internal review.