Jobs at Alice (Formerly ActiveFence)

Lead the rapid development and release of AI safety and security benchmarks. Own evaluation taxonomies, harnesses, quality standards, publication decisions, research timelines, and quarterly roadmap planning. Collaborate with AI labs, universities, internal researchers, and freelance subject-matter experts. The role also requires reviewing research, engaging with frontier labs, presenting findings, and attending industry conferences.

Lead the development and release of AI safety and security benchmarks every two to three weeks. Own evaluation taxonomies, harnesses, quality standards, verification, publication decisions, research timelines, and quarterly roadmap planning. Collaborate with AI labs, universities, internal researchers, and freelancers. The role also requires staying current with AI safety research, maintaining lab relationships, presenting work to clients, and attending conferences.

Lead the development and release of frontier AI safety and security benchmarks every two to three weeks. Own evaluation taxonomies, harnesses, quality standards, publication decisions, research timelines, and quarterly roadmap planning. Collaborate with AI labs, universities, internal researchers, and freelance subject-matter experts. Conduct hands-on evaluation review, maintain ecosystem relationships, present research to clients, and attend conferences.

Principal-level hands-on engineer embedded in a product pod to drive 0-to-1 innovation. Responsibilities include architecting, coding, and launching greenfield products; creating demos and proofs of concept; leading high-stakes technical client discussions; shaping roadmaps and integration specifications; identifying automation opportunities; and resolving critical engineering bottlenecks. The role requires exceptional system design, modern AI expertise, commercial judgment, client presence, autonomy, and experience shipping complex software from scratch.

Serve as the senior technical front line for frontier AI lab clients in the Bay Area. Lead technical discovery, RFPs, solution scoping, client meetings, events, and technical discussions with engineering and research teams. Build the Valley AI and security community, represent the company publicly, translate market signals into research and product strategy, and help establish the company’s regional presence. Partner with sales on technical wins while influencing AI security research, product direction, and local hiring.

Lead post-training evaluation, red-teaming, and reinforcement-learning gym audits for open-weight models. Build benchmarking and evaluation pipelines for LLM security threats such as indirect prompt injection, analyze reward diagnostics and rollout traces, audit agentic environments, produce evaluation cards, and integrate Dockerized environments into distributed training frameworks. The role requires expertise in RL training, AI security, PyTorch, containerization, verifiers, reward shaping, and agentic tool-use workflows.

Own full-cycle global recruitment for specialized technical roles, partnering with executives and hiring managers across regions. Source candidates, optimize AI-driven recruiting workflows, analyze hiring KPIs, support employer branding, and improve hiring-to-onboarding outcomes. The role requires strong communication, data analysis, stakeholder management, international hiring experience, and proficiency with ATS and modern sourcing tools in a fast-paced technology environment.

Lead and grow a multidisciplinary Data Science engineering team building real-time content moderation, GenAI safety and red-teaming systems, and Databricks-based data and MLOps platforms. Set technical direction, own delivery and quality, mentor and hire engineers, review designs, and drive reliability, observability, CI/CD, cost, latency, and production ML excellence across Python, Rust, Spark, AWS, Kubernetes, and model-serving infrastructure.

Conduct Android malware research, reverse engineering, and static and dynamic analysis. Detect emerging threats and attack vectors, analyze applications for malicious behavior, write detection rules, develop automation, and document findings for threat intelligence reports. Collaborate across teams and share technical knowledge. Requires at least three years of relevant research, reverse engineering, or low-level development experience, programming proficiency, and familiarity with malware analysis tools.

Research and analyze fraud detection mechanisms, abuse vectors, platform vulnerabilities, and emerging fraud tactics across web, mobile, social, forum, and dark web environments. Conduct OSINT and WEBINT investigations, validate abuse scenarios, operate technical research infrastructure, analyze datasets and telemetry, automate workflows, monitor third-party channels, and deliver actionable intelligence and mitigation recommendations.

Analyze generative AI content infringements and develop adversarial prompts to identify model vulnerabilities across hate speech, misinformation, intellectual property, and other abuse areas. Manage multilingual datasets, investigate safety circumvention tactics, oversee projects and quality assurance, and collaborate with engineering, product, and policy teams to improve AI safety strategies.

Investigate fraud methods and vulnerabilities across web, mobile, social media, forums, apps, and darknet platforms. Analyze intelligence sources and large datasets to identify malicious activity, behavioral vulnerabilities, and process bypasses. Monitor third-party channels, conduct OSINT investigations, and produce reports with actionable mitigation recommendations. The role may also support red-team exercises, automation, and scripting to uncover abuse methods.

Lead a multidisciplinary red teaming team to design and run adversarial tests, evaluate model risks, produce high-quality deliverables, and communicate findings to clients while improving methodologies and workflows.

Create adversarial prompt-injection tasks in YAML for RL Gym environments: define agent personas, simulated environments, adversarial payloads, benign user requests, run tasks against frontier models, validate success rates, and submit passing tasks with deterministic evaluation criteria.

Investigate and adjudicate complex online child safety cases involving sensitive content, analyze large review datasets for emerging threats and actionable signals, and develop Python or Bash automations to improve review workflows. Produce intelligence reports and policy recommendations, collaborate with engineering and AI teams on labeling quality and golden datasets, and support improvements to automated detection models. The role requires extensive child safety expertise, technical scripting ability, strong communication, and resilience to graphic content.

Analyze content infringements and write adversarial prompts to find model vulnerabilities across LLMs, text-to-image/video, and agents. Manage datasets and projects end-to-end, investigate evasion tactics, collaborate with engineering, product, and policy teams, and promote knowledge sharing to improve model safety.

Conduct in-depth OSINT and WEBINT research to identify online threats. Prepare analytical reports. Collaborate with teams to ensure quality outputs.

The role involves conducting simulated cyber attacks, vulnerability assessments, risk analyses, and developing mitigation strategies for AI security technologies.

Join Alice.io as a GenAI Expert to design adversarial prompts and create diverse prompts that challenge AI systems, ensuring safety insights and cultural nuances. Native language fluency is preferred.

Lead a team of malware researchers, manage security research projects, enhance team reports, and collaborate on threat detection and response.