Key Notes
Rides matching at Bolt is moving from heuristic dispatch to context-aware allocation using machine learning and reinforcement learning. This Principal Product Manager owns the production stack, model roll-out, market experiments and interface between matching and pricing. Relevant expertise includes experience shipping real-time allocation models and maintaining performance during a live-system migration.
What You'll Work On
- Define the dispatch-engine strategy and roadmap across foundational machine-learning investments and near-term matching improvements.
- Coordinate model production from offline training through shadow mode, controlled experiments, market rollout and live monitoring.
- Maintain the performance and stability of the current matching system while its replacement is built.
- Define the signal contract between matching and pricing, including completion-probability and supply-density inputs.
- Design experiments, evaluate algorithmic trade-offs and set priorities for the cross-functional matching pod.
Why This Role Matters
- Bolt’s rides-matching system affects pickup times and driver earnings; the product roadmap and model experiments shape how ride requests are allocated across markets.
- The live dispatch service must remain stable while Bolt introduces probability-based allocation, with shadow deployments, controlled experiments and monitoring testing the new engine before wider rollout.
What They Are Looking For
- Experience: Relevant experience includes work shipping machine-learning models into real-time production allocation systems at scale.
- Domain: Relevant domain knowledge includes experience with matching, dispatch or equivalent systems in a two-sided marketplace, logistics platform or high-throughput environment.
- Experience: Relevant experience includes evidence of maintaining a live system while building and migrating to its replacement.
- Stakeholder: Relevant stakeholder skills include experience coordinating cross-functional product pods and negotiating interfaces with adjacent teams such as pricing or economics.
- Technical: Relevant technical skills include direct exposure to reinforcement learning in production or to teams deploying reinforcement-learning systems.