Lit Financial

Company: GameBrain Inc.

Location: Troy, MI (Year 1 Onsite → Earned Autonomy)

The Mission 

At Gamebrain, we’re applying SOTA computer vision models to help teams dramatically accelerate game prep and scouting, saving days of manual repetitive work and giving coaches and analysts massive leverage over their opponents. One of the biggest unlocks: legal access to real, high-quality sports data through deep relationships in the sports industry. Anyone who has built production CV systems knows how rare, and how valuable, that is.

The GameBrain Philosophy 

We do not want a 50-person engineering org. We want 5 people who are so technically elite and AI-augmented that they outperform 20. We operate on the principle of compounding engineering: you are expected to build tools, auto-labeling pipelines, and data flywheels that make tomorrow’s work exponentially faster than today’s. If you are a Staff Architect who is tired of corporate red tape and just wants to build the most advanced sports-tracking system on the planet, you belong here. You are not expected to know anything about football but you are expected to take ownership of your learning path with enthusiasm and proactiveness in order to ask good questions.

Compensation 

We expect elite talent, and we pay for it. We are fully equipped to compensate at parity with tier-one AV companies (Tesla, Waymo, etc) for the right architect. We do not expect you to take a "startup discount" on your base salary and you will get all the benefits of startup equity.

What You’ll Build

• Design and implement deep learning models, advanced neural networks, and architectures for computer vision problems

• Architect the BEV Pipeline: Design and execute the technical roadmap for automated homography, transforming 2D camera pixel space into accurate 3D field coordinates.

• Pioneer Multi-View Fusion: Design advanced temporal architectures that fuse "Sideline" and "Endzone" views to resolve heavy occlusions and maintain persistent player IDs through the "pile."

• Build the Spatial Engine: Lead the development of 3D Pose Estimation and temporal tracking models for precise player movements.

• Own the Data Flywheel: Architect auto-annotation systems and flexible training pipelines driven by active learning loops. You care just as much about the compounding leverage of your data loop as the architecture of your model.

• Semantic Extraction: Implement VLMs and Transformers to extract semantic data: personnel groupings (11 vs. 12 personnel), jersey recognition, and play-type identification.

The Technical Baseline 

• Elite Credentials: Masters or Ph.D. in Computer Vision, Robotics, AI, or a strictly related field, OR equivalent elite industry experience. A strong portfolio of tier-one publications (CVPR, ICCV, ECCV) or granted patents is nice to have. You must prove that you can ship production-grade CV models.

• The "Spatial" Foundation: Deep, rigorous expertise in projective geometry, camera calibration, and multi-view synthesis. You can explain the math of going from a pixel to a yard line in your sleep.

• Advanced Architecture Mastery: Proven experience designing and deploying multi-task networks that efficiently share backbones across detection, segmentation, and pose estimation.

• AV-Level Rigor: Experience from industries with high-stakes spatial constraints (Autonomous Vehicles, Robotics, Industrial Automation, or high-end Sports Tech).

• Mastery of the Stack: Expert proficiency in Python with PyTorch, TensorFlow or similar frameworks.

• Experience with image and video processing, data augmentation, and working with large-scale datasets

• Familiarity with cloud computing platforms and containerization tools such as GCP, Docker, or Kubernetes is a plus