Uber Technologies Inc. dismissed two senior technology leaders from its newly established artificial intelligence data labeling business, signaling a major organizational restructuring at the division. The Keelung District Prosecutors’ Office was not involved, as this is a corporate personnel move at the San Francisco-based ride-hailing giant. Naga Kasu, Senior Director of Engineering, and Pankaj Kamat, Director of Product at Uber AI Solutions, left the company recently. Both executives spent more than ten years at Uber and previously spearheaded crucial engineering and product operations in its core ridesharing and delivery divisions before leading the launch of the company’s AI data annotation efforts.
A spokesperson for Uber confirmed the departures and stated that the exits are part of a broader leadership transition within the division, which continues to see strong business momentum. The corporate shakeup had a positive impact on Wall Street, with UBER stock rising 1.5% in early open-market trading on Wednesday, July 1, 2026, trading around $73.11. The personnel changes highlight the evolving operational realities of the AI data labeling sector, where managing a massive, global gig-worker workforce for complex machine learning tasks requires a different execution playbook than running a standard ridesharing and food delivery network.
The Operational Architecture of Uber AI Solutions
To understand why this leadership transition is significant, one must examine the unique business model of Uber AI Solutions. Originally launched in late 2024 under the name Scaled Solutions and subsequently rebranded, the division represents a calculated effort by Uber to diversify its revenues and establish a strong presence in the rapidly expanding artificial intelligence infrastructure market. The global business unit is led by Megha Yethadka, who serves as Global Head of Uber AI Solutions.
The core thesis behind Uber AI Solutions is simple: the company is converting its vast, global network of over 8 million gig workers into an on-demand workforce for data labeling, model evaluation, and software localization. Artificial intelligence models require massive volumes of high-quality, human-annotated data to learn logical patterns, recognize objects, and generate accurate outputs. By integrating micro-task capabilities directly into its driver platform, Uber has created an instant, distributed workforce that can annotate data at a scale and speed that traditional data labeling companies find difficult to match.
Turning the Transportation Network into an AI Training Ground
The operational genius of Uber’s approach lies in its ability to utilize driver and courier downtime. During a typical shift, rideshare drivers and food delivery couriers experience periods of inactivity—such as waiting for ride requests during off-peak hours or sitting in parking lots between deliveries. Historically, this downtime represented lost earning potential for the drivers and idle capacity for Uber’s platform.
Uber AI Solutions solves this problem by allowing drivers and couriers to complete “digital tasks” directly within the Uber Driver app during these low-demand periods. These micro-tasks include:
- Image Classification: Labeling specific objects, road signs, and lane markings in pictures to train computer vision models.
- Text Analysis and Translation: Evaluating short text samples and validating translation accuracy across multiple languages.
- Audio Transcription: Listening to short voice clips and transcribing the spoken words into written text.
- Receipt Digitization: Extracting transaction details and line items from photographs of paper receipts to train document-processing algorithms.
- Autonomous Vehicle Video Labeling: Reviewing and labeling video footage to help self-driving cars identify pedestrians, cyclists, and traffic hazards.
By turning these unproductive hours into active earning opportunities, Uber is offering its drivers more choice and flexibility while building a highly competitive, low-cost asset for the enterprise technology market.
The Global Pilots: From 1.4 Million Indian Drivers to US Rollout
To test the viability of this crowdsourced model, Uber has executed a phased rollout of its digital task platform, starting in highly populated developing markets before expanding to Western countries. The company launched its first major pilot project in India in September 2025, enabling drivers across 12 major cities—including Delhi, Mumbai, Bengaluru, Hyderabad, Pune, and Jaipur—to complete data annotation tasks through the app.
The choice of India as the initial testing ground was highly strategic. Uber has over 1.4 million active driver partners in India, providing an immediate, massive labor pool that can perform high-volume data annotation tasks at a fraction of the cost of Western contractors. Following promising early results in India, where drivers completed thousands of tasks within the first few weeks, Uber expanded the pilot program to the United States in October 2025, allowing a select group of domestic drivers and couriers to complete voice-recording and image-uploading tasks directly through the driver app.
Why the Leadership Shakeup Matters for Uber’s AI Playbook
The departure of Naga Kasu and Pankaj Kamat after more than a decade at Uber suggests that the company is transitioning its AI data labeling business from an experimental startup phase to a mature, enterprise-grade operation. Both executives were instrumental in building the initial technical and product foundations of Uber AI Solutions, leveraging their deep experience in ridesharing systems to integrate micro-tasks into the company’s core mobile applications.
However, as the division grows, the operational challenges of managing a global AI training workforce are shifting. In the early stages, the primary challenge was technical integration—building the software hooks inside the Driver app to distribute tasks and process payments. Today, the challenge is quality control, security compliance, and enterprise sales. Large technology companies and AI laboratories are increasingly demanding strict data governance, verified annotator expertise, and robust quality-assurance protocols. The leadership transition suggests that Uber is bringing in new, specialized talent to refine its quality-control algorithms, strengthen its data security standards, and secure long-term contracts with major corporate clients.
The Blue-Chip Client Roster of Uber’s Data Labeling Business
Despite its nascent status, Uber’s data services platform has already attracted several of the world’s most prominent technology and automotive organizations. By offering a massive, globally distributed workforce that can perform multilingual tasks in over 100 languages, Uber AI Solutions has established itself as an indispensable partner for companies building advanced AI systems.
The division’s client list includes several industry-leading organizations:
- Alphabet Inc.: Google’s parent company uses Uber’s data labeling services to help train its advanced search algorithms, machine learning models, and computer vision systems.
- Aurora Innovation Inc.: The autonomous vehicle developer, which famously acquired Uber’s self-driving division (ATG) in 2020, relies heavily on Uber’s driver network to label the complex video and LiDAR footage required to train its driverless trucking software.
- Niantic Inc.: The developer behind the hit augmented-reality game Pokémon Go uses Uber’s crowdsourced image data to map real-world environments and build advanced 3D spatial models.
- Luma AI: The generative video startup utilizes Uber’s global task network to evaluate and label video clips, helping to train its next-generation text-to-video algorithms.
- Tier IV: The Japanese self-driving software developer partners with Uber AI Solutions to obtain high-quality, localized annotations for its autonomous driving platforms.
Furthermore, industry sources indicate that Uber is actively pursuing other major enterprise customers, including cloud software giant Salesforce Inc., to expand its client roster and secure its position as a primary supplier of AI training data.
The High-Stakes Battle Against Scale AI and Appen
By entering the data services market, Uber is positioning itself as a direct competitor to established, multi-billion-dollar data labeling platforms. The move represents a major threat to traditional providers, as Uber can leverage its existing, massive user base of 8 million global gig workers without needing to spend millions of dollars on user acquisition or marketing.
This entry has disrupted a highly lucrative, rapidly growing industry. The global data labeling market is valued at approximately $2.32 billion in 2026 and is projected to reach over $14.9 billion by 2034, registering a compound annual growth rate of over 22%. To defend their market share against Uber’s low-cost crowdsourced model, incumbent players are taking drastic strategic actions. For instance, industry leader Scale AI, which secured a massive valuation after a $14.3 billion investment from Meta, is shifting its focus toward high-end, expert-driven human-in-the-loop (HITL) evaluations, while traditional providers like Appen are struggling to adapt to the changing competitive landscape.
The Hyper-Growth of the $2.32 Billion Data Labeling Market
The rapid expansion of the data labeling market is a direct result of the explosive growth of generative artificial intelligence and large language models. While early machine learning models relied on simple, automated data classification, modern generative AI systems require highly complex, nuanced human feedback to ensure safety, accuracy, and alignment with human values.
This demand has created an urgent need for specialized training data, with global enterprise spending on AI systems projected to reach hundreds of billions of dollars over the next few years. Because the performance of an AI model depends heavily on the quality of its training data, companies are willing to pay a premium for high-quality, human-annotated datasets. This massive investment wave has created significant room for growth, prompting non-traditional tech companies like Uber to enter the market to capture a share of the infrastructure spend.
Controlling Pricing and the Gig Worker Pay Controversy
While Uber’s data labeling venture offers clear strategic advantages, it has also introduced significant labor and operational friction. In many ways, Uber’s digital task network resembles traditional micro-task hubs like Amazon Mechanical Turk, where independent gig workers complete small, repetitive online tasks for very low pay.
This model has faced growing criticism from contractors and labor advocacy groups, who accuse the platform of labor exploitation:
- Mid-Project Pay Cuts: Contractors on active online forums like Reddit have reported significant, unannounced pay cuts mid-project, with rates for video evaluation tasks dropping from $6.25 per task to $4.12, and eventually down to $2.08.
- Extremely Low Hourly Rates: Many contractors complain that after accounting for the time required to complete complex assessments and language tests, their effective earnings drop to a fraction of the minimum wage, sometimes averaging just $1.09 per hour.
- Lack of Accountability: Workers have reported complete silence from support teams when disputing unpaid tasks, raising serious questions about the platform’s payment transparency and quality control.
If these labor disputes continue to escalate, they could lead to regulatory scrutiny from agencies like the Federal Trade Commission (FTC), which has increasingly focused on protecting gig workers from predatory pricing and deceptive wage promises in the digital economy.
Sovereignty, Compliance, and the Enterprise AI Standard
To transition from a basic micro-task hub to a trusted enterprise data partner, Uber AI Solutions must address strict security and compliance standards. Large enterprise clients, particularly those in highly regulated industries like finance, healthcare, and autonomous driving, cannot risk exposing their proprietary data to anonymous, unvetted gig workers.
Consequently, the advanced data labeling market has established a strict set of compliance baselines:
- Security Certifications: Providers must hold robust industry certifications, including SOC 2 Type II, ISO 27001, and HIPAA compliance, to guarantee that sensitive data is handled securely.
- Data Residency and GDPR: Under strict European regulations like the General Data Protection Regulation (GDPR), companies must ensure that personal data is processed within secure, legally compliant jurisdictions, which is difficult to manage with a completely decentralized, global contract workforce.
- Layered Review Hierarchies: To ensure high data quality, leading platforms like Scale AI and Appen utilize structured, multi-layered quality-assurance protocols, where expert, named analysts review and validate the work of frontline annotators.
For Uber, building a platform that can meet these rigorous enterprise standards will require significant engineering and operational investments. The recent leadership shakeup at Uber AI Solutions suggests that the company is reshuffling its product and engineering teams to address these high-level compliance and quality-control challenges, aiming to transform its raw gig-worker network into a secure, professional, and compliant data engine.
Conclusion
The sudden dismissal of Naga Kasu and Pankaj Kamat from Uber AI Solutions represents a critical pivot point in the company’s ambitious plan to become the “Uber” of the data services market. By reshuffling its product and engineering leadership, the company is attempting to transition its nascent data labeling division from a basic micro-task pilot into a highly sophisticated, enterprise-grade data engine. Leveraging its massive global network of 8 million gig workers, Uber has built a highly disruptive, low-cost platform that poses a direct challenge to industry giants like Scale AI and Appen.
However, the path to long-term success is filled with operational and ethical challenges. To win the trust of lucrative enterprise clients like Google and Salesforce, the company must resolve ongoing issues with data quality, security compliance, and international data residency regulations. Furthermore, the company must address growing labor friction and contractor complaints regarding mid-project pay cuts to avoid regulatory backlash and ensure a sustainable workforce. As the global AI market continues to expand, Uber’s ability to balance low-cost crowdsourcing with strict enterprise quality standards will determine whether its data labeling division can become a major, multi-billion-dollar growth engine for the transportation giant.
