Senior Software Engineer, ML Platform

Stack AV•Pittsburgh, PA

About The Position

Stack is developing AI and autonomous systems for the trucking transportation industry, combining artificial intelligence, robotics, machine learning, and cloud technologies. The ML Data Understanding team provides trusted, useful data for Stack’s machine-learning applications across data mining, training, and safety evaluation, partnering with autonomy teams on data engineering, mining, modeling, and infrastructure. This role helps set the direction for and build data-mining and semantic-search services, including inference services that use large language models, open-world models, and vector databases. It also contributes to infrastructure for machine-learning training and inference workloads, using open-source components such as Ray, Spark, Lance, and Iceberg. The position spans onboard C++ systems and offboard machine-learning and data infrastructure, supporting data discovery and development of autonomous-driving products.

Requirements

Experience with both machine-learning platforms and building machine-learning-based applications.
At least 5 years of relevant domain experience.
Proven track record of building scalable, reliable infrastructure in a fast-paced environment.
Deep understanding of design trade-offs and ability to explain them and achieve alignment with others.
Hands-on experience developing semantic-search infrastructure, generating multimodal video, image, or audio embeddings, and managing vector databases.
Experience with multimodal data indexing, model-training infrastructure, model optimization, or large-data-processing pipelines.

Nice To Haves

Experience building machine-learning models or infrastructure in autonomous vehicles, perception, or decision-making domains.

Responsibilities

Architect multimodal semantic-search and data-discovery platforms for autonomous-driving development, using AI-assisted software engineering.
Investigate and apply large language models, visual language models, and Retrieval Augmented Generation techniques.
Build agentic data-mining, semantic-search, and data-curation workflows as part of a recursive machine-learning improvement loop.
Develop data-understanding infrastructure for real-time querying, vector databases, and batch or stream processing using technologies such as Ray, Spark, and LanceDB or similar tools.
Deliver end-to-end data-mining solutions across onboard C++ and offboard machine-learning and data infrastructure.
Develop end-to-end real-time semantic-search services for text, images, and videos, including vector databases.
Identify issues in existing machine-learning infrastructure and proactively improve system performance.
Build low-latency, high-throughput batch or stream-processing pipelines.
Drive technical discussions across organizations and deliver solutions on time.
Write readable, high-performance Python and C++ code.