SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Mapbox is the leading real-time location platform serving over 4 million registered developers. This Senior AI Engineer role is scoped by technical skill rather than a single product, spanning Search and Places data, Location and Navigation Intelligence, and Platform work that enables AI agents to use Mapbox.
You will own the technical design and delivery of multi-component AI systems, accountable for quality and production performance. Key responsibilities include:
- Define product behavior, build measurement frameworks, and create evaluation systems for non-deterministic AI behavior. Formulate and validate hypotheses with appropriate datasets and regression testing.
- Run continuous evaluation of APIs, SDKs, data representations, and reference applications from end-user perspectives (developers, agents, consumers). Identify integration anti-patterns and drive fixes.
- Own MVP delivery against agreed technical designs, balancing perfection with shipping useful increments.
- Build data pipelines: ingestion, conflation, entity resolution, quality checks, batch and streaming jobs for large datasets.
- Track and evaluate external datasets, models, and benchmarks from research and open-source; decide build vs. buy.
- Design feedback loops where product usage generates improvement data. Instrument systems for reproducible failures.
- Design boundaries between models and their tool calls. Build model harnesses, decide delegation, maintain state consistency.
- Optimize for latency and cost per request: streaming, partial results, caching, model routing, prompt structure.
- Build internal tools (CLI, MCP, etc.) for team iteration and share generalizable components.
- Raise team bar through code/design review and evaluation practice mentorship.
- Participate in on-call rotation for 24/7 system availability, including potential off-hours response.
REQUIREMENTS:
- Bachelor's degree in STEM discipline and 5+ years software engineering experience with production ownership of services, pipelines, or SDKs.
- 2+ years shipping LLM-backed features to real users in systems with error budgets, on-call rotations, and customer-facing regressions.
- Data engineering depth: SQL, at least one distributed processing framework, experience with pipelines where data quality matters more than speed.
- Fluency with tool calling and agent orchestration, including failure modes (stale context, hallucinated arguments, silent partial success, unbounded loops).
- Strong Python or TypeScript; comfort reading code in any caller language.
- Direct experience or deep understanding designing evaluations for non-deterministic systems; can describe a dataset built and failure caught.
- Working knowledge of multiple agent harnesses with opinions on their weaknesses.
- Experience diagnosing latency in distributed request paths.
- Comfort with ambiguity and judgment to ship narrow working solutions while general solutions remain unclear.
NICE TO HAVE:
- Geospatial data experience (routing, geocoding, POI/address data, OpenStreetMap, conflation).
- Public API or SDK design for developers you never meet.
- MCP or similar tool transport experience.
- CI eval running with commercial or custom harnesses.
- Automotive, in-vehicle infotainment, CarPlay, or Android Auto experience.
- Voice pipeline experience (streaming ASR, TTS, barge-in, endpointing, wake word).
- Constrained compute, offline, or intermittent connectivity experience.
- Product launch through first external integrations where customers find gaps.
