Senior ML Engineering Manager (AV data)

About the job

Autonomous vehicles are coming to the markets in which inDrive serves, but the AV industry has a blind spot: it has almost no data from the cities where we operate. inDrive runs one of the world's largest ride-hailing fleets, concentrated in exactly the emerging markets global AV players have yet to reach. We are building the team that turns that fleet into an AV grade data platform: large scale multi sensor collection from real vehicles on real trips, evaluation of driving models against human drivers, automated and human in the loop causal annotation, and camera first 3D scene reconstruction for simulation.

We are hiring the founding leader of this workstream. This is a rare true zero to one: you will hire the team, stand up the vehicles and the data platform, set the technical bar, and own delivery end to end.

• Own the AV data workstream end to end. Roadmap, execution, quality, and outcomes, from vehicle data capture through processing, annotation, and delivery of AV-grade datasets
• Build the team from zero. Hire and lead a multi disciplinary team covering large scale data engineering, search and serving, ML engineering across divergence detection, edge inference, and VLM based auto labeling, 3D reconstruction research, and data annotation operations. Own the hiring bar and the team's growth
• Management: Manage ML engineers' charter across the cluster
• Set technical direction across disciplines. Provide design authority for the full pipeline, including onboard capture, cloud ingestion and processing at fleet scale, semantic and causal labeling, and neural scene reconstruction, and make the key technical decisions on evidence
• Partner with leaders across the organization. Work closely with the leaders of data platform, ML ops, geo and mapping, and inDrive's wider autonomous vehicle initiatives to align roadmaps, share infrastructure, and unblock delivery
• Own external partnerships. Drive the technical relationships this work depends on, including the NVIDIA AV ecosystem and prospective AV partners consuming our data
• Balance research and delivery. Several of the core questions here are open research. You frame milestones as decision gates, redirect work based on evidence, and keep the team shipping measurable results on a schedule

• 10+ years of engineering experience in applied ML, autonomous systems, or large-scale data and ML platforms, including 3+ years in the autonomous vehicle industry or closely adjacent autonomy domains (AV data, simulation, or evaluation platforms, ADAS, robotics perception, HD mapping), and 4+ years managing engineering teams, with a track record of building teams from scratch and delivering from zero to one
• Strong command of the modern AV and foundation model landscape: vision-language models, multi-modal perception, 3D reconstruction and neural rendering, and the NVIDIA AV ecosystem. Deep enough to set direction, review designs, and make evidence-based technical calls
• Genuine engineering leadership range: you have successfully led data engineering, platform, or infrastructure delivery, not only ML research, and can evaluate and unblock work across those disciplines
• Proven ability to take large-scale ML systems from conception to production: data pipelines, training and evaluation infrastructure, and models running under real-world constraints
• Strong external presence: experience owning technical partnerships, vendor and licensing relationships, or customer-facing technical engagements at a senior level
• Deep fluency in the modern ML stack (PyTorch or equivalent, large-scale training and inference, Python and Linux) and in cloud data platforms at multi-terabyte scale
• Bachelor's degree in Computer Science, Engineering, or a related field

Nice to have:

• AV industry experience specifically in offline evaluation, data mining, annotation platforms, or simulation, as distinct from on-vehicle stack development
• Advanced degree (MS or PhD) in ML, robotics, or computer vision
• Experience operating in emerging markets, and comfort building where the playbook does not exist yet