Senior Architect (AI – Computer Vision & Sports Intelligence Platform)

Company Description

str8bat is a deep-tech sports technology company building the next generation of sports intelligence infrastructure. Starting with cricket, we convert broadcast video, motion data, and contextual match information into real-time performance intelligence and storytelling for broadcasters, teams, and fans.

Our mission is simple: help players Play Better by turning motion into actionable insight. This role sits at the intersection of AI, computer vision, real-time systems, broadcasting, and human-motion intelligence — and very few teams anywhere are solving it at this depth.

Role Overview

We are hiring a deeply technical, hands-on Senior Architect to design and build str8bat's Sports Intelligence Platform end to end — and lead from the front. You will own the architecture spanning computer vision, AI orchestration, real-time video, and event-driven systems, translating an ambitious product vision into production systems that hold up under live broadcast conditions.

Your mission: take a broadcaster's live video feed, reconstruct the bat path and its descriptors, compare them against str8bat's library of ~2–3 million sensor-captured motion records, and estimate sensor-grade metrics — bat speed, impact speed, and a timing index — then fuse bat path with body and weight-transfer cues into deeper motion intelligence.

Key Responsibilities

• Reconstruct bat path and bat trail from a broadcaster's video feed — pose estimation, bat tracking, temporal tracking — robust to zoom changes, camera movement, inconsistent framing, motion blur, and partial occlusion.

• Design the descriptor layer that turns each reconstructed bat path into a structured, comparable representation (geometry, velocity profile, impact point, timing).

• Build the comparison-and-estimation engine that aligns those descriptors against str8bat's library of ~2–3 million sensor-captured motion records to estimate sensor-grade metrics — bat speed, impact speed, and timing index — none of which are directly measurable from the video.

• Fuse bat-path signals with body pose and weight-transfer cues to produce higher-order motion intelligence and shot-quality insight.

• Extend the intelligence layer into player-baseline comparison, contextual prediction, temporal trends, and pressure-based performance analysis.

• Own the end-to-end, event-driven platform architecture, built around canonical sports event objects, and the synchronization layer across video, metadata, scoring, and motion.

• Define the near-real-time processing infrastructure and scalable APIs that broadcasters and downstream systems consume.

• Design scalable inference pipelines for live / near-live deployment, optimized for latency and production robustness; architect vector search, embeddings, temporal sequence modeling, retrieval, and hybrid AI + rules-based systems that scale across sports.

• Lead technical execution across the stack, orchestrate agentic, AI-assisted development workflows, and build and mentor a lean, high-leverage engineering team.

• Partner with product and broadcasters to turn live-production constraints into systems consumable by broadcast graphics, commentary, OTT, and fan-engagement platforms.

Required Technical Skills

• Strong systems-architecture fundamentals, with the range to work across both research and production.

• Deep computer-vision and ML expertise: Python, PyTorch / TensorFlow, OpenCV, video-processing systems, and real-time inference pipelines.

• Event-driven, streaming, and distributed systems; GPU inference optimization; vector databases and retrieval systems.

• Hands-on with pose estimation, object tracking, temporal modeling, embeddings and similarity search, and video analytics.

• Cross-modal alignment / domain adaptation — relating video-inferred motion to sensor (IMU) data, and estimating physical quantities from indirect or weakly-supervised visual input.

• Sharp debugging instincts and a first-principles problem-solving mindset.

Strong Plus

• Sports analytics, broadcast / media-tech, or multi-camera synchronization experience.

• AI-assisted / agentic engineering workflows.

• Building scalable production AI infrastructure, and working with large-scale behavioral or motion datasets.

Engineering Mindset We Value

• Systems thinking over isolated model-building.

• Ability to simplify complexity and own outcomes end to end.

• Speed of execution and comfort operating in ambiguity.

• Deep curiosity about human movement and performance.

• Pragmatism over academic perfection.

Target Candidate Profiles

• CV / ML architect who has shipped real-time video-understanding systems to production.

• Markerless motion-capture or sports-biomechanics engineer comfortable with 3D human and object pose from monocular video.

• Staff / Principal engineer from a deep-tech, sports, or media-tech startup who has owned platform architecture end to end.

• ML systems engineer strong in retrieval, embeddings, and inference at scale.

• Hands-on engineering leader who has built and run lean, high-leverage teams.

Why This Role Matters

This is one of the first event-centric sports intelligence platforms — and the opportunity reaches far beyond cricket analytics. Combining motion understanding, contextual intelligence, and historical correlation can become foundational infrastructure for broadcasting, coaching, performance analysis, fan engagement, and multi-sport human-motion intelligence.

Role Details

Location: Bangalore

Employment type: Full-time employee

Reports to: CEO