About the role
At Lyft, our purpose is to serve and connect. We aim to achieve this by cultivating a work environment where all team members belong and have the opportunity to thrive.
Marketplace teams are at the heart of Lyft’s products and decision-making. We’re looking for passionate, driven engineers to build systems that empower our users (both drivers and riders) to make the most effective use of Lyft’s products and experiences by making them more predictive, personalized, and adaptive. The ideal candidate enjoys solving problems with distributed computing for ML and forecasting models, and finding patterns and insights in real-world data while building reliable systems, and is excited about working in a fast-paced, innovative, and collegial environment.
The Forecasting and Real-Time Optimization Platform (FORTOP) team in Lyft’s Rideshare Experience & Marketplace (REM) org provides reliable, real-time market supply and demand signals and forecasts that power the systems making critical automated decisions for Lyft’s business. These signals feed many of Lyft’s most important marketplace products, including Dynamic Pricing, Real-Time Supply Management, Fulfillment, and more.
As a Senior Software Engineer on FORTOP’s forecasting team, you’ll lead the design and operation of the production systems that turn streaming marketplace data into accurate, reliable forecasts: real-time feature pipelines, model training and inference infrastructure, backtesting and evaluation tooling, and the serving layer that downstream products query continuously. You’ll work closely with our scientists to take models from prototype to production, and with downstream teams to define the contracts and reliability guarantees they build on. The work involves forecasting both rider demand and driver supply as they shift with traffic, holidays, and major events, and keeping systems correct and fast during the peaks when they matter most.
Responsibilities
- Lead projects end to end: Drive multi-engineer projects from proposal and technical design through launch. Break work into milestones, align dependencies with partner teams, and own estimates.
- Architect ML systems: Design and evolve streaming feature pipelines, training and inference workflows, and signal serving, and lead design reviews with attention to latency, correctness, scalability, and cost. Reduce complexity by consolidating or sunsetting components.
- Partner with science to ship: Partner with scientists to design/productionize ML models. Build the rollout tooling (shadow, canary, rollback) and the backtesting and experimentation infrastructure that make launches safe and fast.
- Own signal contracts: Define and maintain the interfaces, data contracts, and SLAs between forecasting signals and downstream consumers such as Dynamic Pricing and Driver Incentives.