Machine Learning Manager
About the job
- An undergraduate or graduate degree in computer science or a similar technical field.
- At least 4 years of experience as a machine learning engineer, including training large deep learning models and working with real-world data.
- Knowledge of at least one machine learning focus area, such as computer vision, audio, or natural language processing.
- At least 2 years of experience managing machine learning teams.
- Ability to understand and make well-reasoned tradeoffs when designing features.
- Ability to set a roadmap and goals for a team and its individual members, delegate, mentor, and deliver results.
- Willingness to interview engineers, collaborate with a recruiting team, and onboard new team members.
- Experience collaborating with product managers and the labeling team to deliver model improvements.
- Interface closely with product management, engineering, DevOps, labeling, and sales teams to build a roadmap supporting the team's long-term vision.
- Lead a team of machine learning engineers and help them achieve their goals through mentorship.
- Participate in product technical design and architecture.
- Participate in the full development cycle, including data collection, labeling, model development, experimentation, training, testing, and production deployment.
- Drive delivery for product milestones, continually release models with new, well-tested features, and ensure quality metrics are achieved.
- Implement and manage security protocols, including training, code review, and best practices.
- Own and manage the risk and security of the business function in coordination with the Security Team.
- Maintain awareness of industry best practices for data maintenance and handling as they relate to the role.
- Adhere to policies, guidelines, and procedures for protecting information assets.
- Report actual or suspected security or policy violations and breaches to an appropriate authority.
About the company
Hive builds AI models for businesses that need to understand, search, and generate digital content, with a focus on online platforms and media companies. Its content moderation tools classify harmful material in images, video, text, and audio, while separate models identify synthetic media and impersonation. Developers connect the models to their applications through cloud APIs, and teams use dashboards to review flagged content and manage decisions. Customers pay for the amount of content processed, with custom pricing and support for enterprise deployments; the platform also runs image search and content generation models. Founded in 2013, Hive is headquartered in San Francisco.
Company Size
201-500
Company Stage
Series D
Total Funding
$142.2M
Headquarters
San Francisco, California
Founded
2013
Get referred to Hive
See people who can refer or advise you
Simplify's Take
What believers are saying
- Hive reports production scale: its customers send billions of API requests each month.
- Early-career engineers get broad ownership: Hive's ML role accepts 1-2 years' experience and spans model design through deployment.
- Developer trials lower adoption barriers: Hive gives $50-plus credits after adding payment details and supports pay-as-you-go access.
What critics are saying
- Self-service adoption has a ceiling: Hive caps visual moderation and AI image detection at 100 daily requests.
- Child-safety technology depends on partners: Hive's CSAM service uses Thorn's classifiers and regularly updated hash database.
- Expect urgent customer work on-site: Hive's San Francisco forward deployed engineer role owns technical support tickets.
What makes Hive unique
- Hive owns a large labeling operation: it reports over five million contributors supporting its training data.
- Hive links detection to reporting: its child-safety dashboard submits flagged material to NCMEC's CyberTipline.
- Hive extends beyond safety: sports teams use its AI to benchmark sponsorship prices and measure brand exposure.
Help us improve and share your feedback! Did you find this helpful?
Benefits
Competitive Pay
Equity
Comprehensive Insurance
Catered Meals
Corporate Gym Membership
Growth & Insights and Company News
6 month growth
0%1 year growth
1%2 year growth
2%Introducing Celebrity Voice Detection Model. Hive. August 6, 2026 Voice is one of the most personal and recognizable parts of a public figure's identity, and generative AI has made it easier to capture and replicate. As generative tools become more accessible, platforms face a growing challenge: identifying when media uses, imitates, or is generated from a recognizable person's voice before it misleads audiences, violates consent, or creates reputational and safety risks. With voice, the harm can be immediate. A realistic clone can make it sound like a celebrity, politician, athlete, or other public figure said something they never actually said. That's why Hive is excited to launch Celebrity Voice Detection, a new AI model that is part of its broader IP & Publicity Detection suite, built to help generative AI companies and digital platforms identify when audio includes or recreates the voice of a well-known public figure. How content moves through Celebrity Voice Detection Powered by its AI technology, Celebrity Voice Detection detects over 500 celebrities, politicians, and other well-known public figures. Comparing voices detected in content against its regularly updated celebrity index, Celebrity Voice Detection returns each identified match and associated confidence scores. Additionally, Celebrity Voice Detection returns an attribute category, which describes what type of famous figure the recognized individual is. The available categories are: film, music, media, politics, business, and sports. This granular metadata can be applied across audio generation and content review workflows, helping platforms review uploaded content and generated clips before they are published. Together, these capabilities provide platforms with stronger safeguards against potentially unauthorized celebrity voice replicas, while preserving space for policy-compliant AI voice creation. AI-generated audio is spreading fast While its industry-leading AI-Generated Audio Detection Model determines whether audio was generated or manipulated by AI, platforms increasingly need to identify if audio contains the vocal likeness of public figures. Synthetic audio is making voice impersonation harder for platforms to ignore. When AI-generated voices replicate people like Morgan Freeman, Matthew McConaughey, Scarlett Johansson, or Martin Luther King Jr., the issue is no longer just whether the audio is real, but whether someone's actual voice or vocal likeness are being replicated without consent. Proposed legislation like the NO FAKES Act, which Hive has endorsed, reflects this broader shift toward protecting individuals from unauthorized AI-generated replicas. Hive has paid close attention to what companies need in this environment, which is why Hive created its Celebrity Voice Detection model. Paired with its AI-Generated Audio Detection Model, platforms can add a more targeted layer of review to AI voice generation. Not only can platforms detect whether a clip is synthetic, but teams can now prioritize higher-risk content that contains a public figure's voice. Get started with Celebrity Voice Detection To learn more about how Celebrity Voice Detection works, explore the technical documentation. If you'd like to test the model, please reach out to [email protected] or contact its sales team and its team can share API keys and credentials for your new endpoints.
Introducing Hive Detect for Enterprises. Hive. Today, Hive is excited to launch Hive Detect for Enterprises, a new way for organizations to access Hive's best-in-class AI-generated and deepfake detection models in a team-ready user experience. This enterprise application delivers a streamlined, drag-and-drop workflow that enables teams to quickly verify the authenticity of images, videos, music, and speech without using an API. Designed for organizational verification. For many teams, verifying content needs to happen quickly and without technical setup. Hive Detect for Enterprises is built for workflows where people need to upload a file, get an answer, and make a decision without using an API or writing code. This approach supports a wide range of use cases, from newsrooms and public sector teams to commercial organizations that need an easy way for employees to check content as part of their day-to-day work. Enterprise access also makes it possible to use detection results in commercial settings, with the structure and limits required for ongoing use. A team-ready platform. Hive Detect for Enterprises takes the drag-and-drop Detect experience and makes it usable for company teams. It adds authentication, access controls, centralized user management, and shared task history, along with higher rate limits and support for longer media uploads. Teams can upload content and receive detection results immediately. For supported media, results are shown frame by frame with clear color indicators that reflect the likelihood of AI generation. Only relevant data is shown for each file type, which keeps results easy to read and interpret over time. For those who want to explore the technology first, a free demo is available at https://hivedetect.ai/. Continuing its enterprise commitment. Hive Detect for Enterprises represents the next step in the availability of Hive's industry-leading AI-content detection technology. If your organization needs a fast, scalable, and intuitive way to verify AI-generated content, Hive Detect for Enterprises is built for you. Please reach out to its sales team at [email protected] or contact Hive here for further questions.
Introducing new free X bot to analyze and detect AI-Generated content for all users. Hive is excited to announce the launch of a new bot on X that uses its industry-leading AI models to analyze and share results in real time, completely free for users. How it works. Anyone on X can simply tag @hive_ai and ask whether a post, image, video, or audio clip is AI-generated. There's plenty of flexibility in how you phrase your question, the bot understands a wide range of prompts, such as: * Is this AI-generated? * Is this video genuine? The audio sounds AI generated. * Is this real or is this another AI-generated photo? Hive's detection models will automatically analyze the media and reply in real time with the results, providing them directly in-thread. In the reply, Hive provides confidence scores for whether the input is likely to be AI-Generated or Deepfake. Videos and audio files will also return frame-by-frame analysis. Finally, Hive identifies probabilities for which generative engines likely created the content (such as Sora2, GPT, etc.). Accessible AI detection for X users. As AI-generated and deepfake content becomes harder to distinguish from reality, tools like this bot are essential for restoring trust and transparency online. Every day, manipulated media spreads across social platforms, making it easy for misinformation to take hold. By making detection accessible to everyone, Hive is helping rebuild confidence in the content Hive see and share. Beyond that, this launch marks an important step in bringing Hive's enterprise-grade detection technology to everyday users. Best-in-Class technology. Today, Hive's industry-leading AI-generated and deepfake content detection technology is trusted across both the public and private sectors. In 2024, an independent research study identified Hive as the "clear winner" in a study that found its AI-generated image and video detection model outperformed competing models as well as human expert analysis. Its technology was also selected among 36 competing solutions for a Department of War contract to support the U.S. Intelligence Community for deepfake detection of video, image, and audio content. More recently, the Department of Homeland Security's Cyber Crimes Center has deployed Hive's AI-Generated and Deepfake Detection technology to support its investigations. With this bot, Hive is giving all users the power to verify what's real. Try it out by tagging @hive_ai on X. You can upload individual media files to check for ai-generation and deepfake content at https://hivedetect.ai. Learn more about its enterprise AI models here.
Virgin Money has collaborated with smart home energy specialist Hive to launch The Retrofit Boost, a mortgage product designed to help customers improve their property's energy efficiency.
BOT or NOT? This special series explores the evolving relationship between humans and machines, examining the ways that robots, artificial intelligence and automation are impacting our work and lives.TrueMedia launched its deepfake detector on Tuesday morning. (TrueMedia image)Seattle-based nonprofit TrueMedia released a free AI-powered media verification tool Tuesday to help journalists and fact-checkers detect deepfakes and combat misinformation ahead of upcoming U.S. and international elections.The non-partisan organization, which launched in January, is led by Oren Etzioni, University of Washington professor and longtime AI specialist, and funded by Uber co-founder Garrett Camp through his Camp.org nonprofit foundation.Oren Etzioni is founder of TrueMedia. (Allen Institute for AI Photo)Although the tool isn’t perfect, its ability to identify deepfakes is “extremely high,” with about 90% accuracy across images, video, and audio, Etzioni said. TrueMedia uses a combination of internally developed technology and AI detection tools from its partners to analyze media and come up with a probability that content is fake.For example, the tool automatically labeled as “highly suspicious” a known fake video that purported to show Ukraine’s top security official claiming responsibility for the March 22 terrorist attack at a Russian concert hall. The tool stated with 100% confidence that the video contained AI-generated imagery.“If it’s a deepfake, we’re very likely to catch it,” Etzioni said.In addition to launching the new tool Tuesday morning, TrueMedia reached a memorandum of understanding with Microsoft to share data and resources, collaborating on different AI models and approaches.Other partners of TrueMedia include Hive, Clarity, Reality Defender, OctoAI, AIorNot.com, and Sensity.The New York Times covered the launch of the tool Tuesday, citing examples including a fake image of Etzioni in the hospital that he generated using an AI tool