À propos
In this role, you'll design, build, and optimize the full player-classification pipeline - detection, tracking, team assignment, jersey-number recognition, and role inference - while delivering the AI-enabled video initiative that turns commodity camera feeds into data-rich experiences: live stats, real-time highlights, and connected-app integrations that delight fans and power coaching tools.
The ideal candidate combines hands-on ML engineering expertise with deep computer-vision knowledge and a passion for sports. You're excited by the challenge of solving hard domain problems - single-camera far-field footage, low-resolution jersey numbers, extreme occlusion, and thousands of unseen uniform configurations - while shipping models that operate reliably at scale.
The outcomes you'll deliver
• Real-time score extraction: Deliver production-grade score data (time, scores, period/inning) from live video with no more than two minutes of latency, reaching alpha within one month and scale-readiness within six months.
• Statistics and player-tracking data: Build models that extract per-player and per-team statistics from video across football, basketball, baseball, volleyball, and soccer - with statistics at scale in nine months and player-tracking data at scale in twelve months.
• Player-classification pipeline: Deliver a robust, end-to-end system for player detection, multi-object tracking, team assignment, and jersey-number recognition that generalizes to unseen schools and uniforms at inference time.
• Build/buy/partner recommendations: Systematically evaluate internal models against external vendor solutions on speed, quality, and price, and deliver clear recommendations on where to build, buy, or partner.
• Fan engagement and retention: Power personalized highlights, live in-app stats, and coaching tools that move the KPIs that matter - session time, retention rates, and new subscriptions.
In This Role, You Can Expect To
Own the player-classification pipeline end-to-end: train and deploy detection (YOLO-family), tracking (DeepSORT-style), team-assignment, and jersey-number recognition models optimized for single-camera, far-field high school footage. Develop and maintain a video-tokenization model that maps video tokens to a sports taxonomy and ontology, delivering contextual outputs (scores, statistics, player tracking) optimized for venue-edge, cloud, and client deployment. Engineer scalable real-time pipelines that connect live video streams and scoring feeds to content outputs, statistics APIs, and our Unified Data Store (UDS) with sub-two-minute latency. Solve domain-generalization challenges: build models that perform reliably on unseen schools, uniforms, camera setups, and lighting conditions using techniques such as test-time adaptation, meta-learning, and contrastive pretraining. Design and run structured evaluations of external partner and vendor CV solutions, benchmarking against internal models on speed, accuracy, sport coverage, and cost to inform build/buy/partner decisions. Apply multimodal fusion techniques - combining video with audio, scoreboard overlays, rosters, and play-by-play - to improve extraction accuracy and enable richer fan experiences. Collaborate with Data Platform, BI/Analytics, Product, and academic research partners (including university collaborations) to advance PlayOn's computer-vision capabilities and translate research into production. Monitor and optimize model performance for latency, accuracy, and cost efficiency across thousands of concurrent streams, using school-disjoint evaluation splits and difficulty-stratified test sets. To Thrive In This Role, You Have
4+ years in AI/ML engineering or applied research, with strong Python and ML frameworks (PyTorch, TensorFlow, or similar). Demonstrated experience with object detection and multi-object tracking pipelines (e.g., YOLO-family detectors, DeepSORT/ByteTrack) in video domains. Hands-on work with at least two of the following: jersey/number recognition or OCR at low resolution, team/appearance classification from crops, domain generalization or zero-shot transfer, or temporal aggregation across video frames. Familiarity with real-time data pipelines (Kafka/Confluent Cloud, Flink SQL, or Spark Streaming) and large-scale cloud infrastructure (AWS S3, Lambda, GPU instances, Snowflake). Strong skills in experimental design, model evaluation (mAP, IDF1/HOTA, per-class accuracy), and rigorous A/B testing against production baselines. Experience evaluating third-party AI/CV vendors or managing build-vs-buy technical assessments. Bonus: knowledge of sports analytics, prior work with multimodal fusion (video + audio + structured data), academic research collaboration, active learning or efficient annotation strategies, or experience optimizing models for edge deployment. How You Play
Ownership over Participation
- You take responsibility for achieving holistic outcomes, prioritize key objectives, and adapt quickly when situations require a different approach. You follow through even against the toughest challenges. Team over Stars
- You are a bridge builder, establishing processes and relationships with teams outside your own. You rally around common goals, find win-win solutions, compromise when necessary, and help others succeed. Growth over Comfort
- You are driven by a desire to grow and actively seek opportunities to expand your comfort zone, skills, and confidence. You embrace new challenges with curiosity, accepting discomfort and failure as opportunities to learn. Fairness over Popularity
- You approach decisions with a scientist's mindset, challenging assumptions and remaining objective. You consider long-term impact over short-term gains, proactively seek others' perspectives, and manage emotions in decision-making.
Company Overview
PlayOn is a dynamic growth-stage company dedicated to championing the spirit of play in the high school space. Backed by KKR, our family of brands-including GoFan, NFHS Network, and MaxPreps-empowers schools with innovative solutions and exceptional service. Our fan engagement platform is the only one that offers event ticketing, streaming, fundraising, concessions, merchandise sales, and website management in one place. We save administrators time so they can focus on what truly matters: supporting the students, staff, and fans who bring their programs to life.
Trusted by thousands of schools across the country, we're here to help create more instant replays, hold-your-breath moments, last-minute comebacks, and games you want to watch over and over again.
When being there means everything, we make sure you never miss a moment.
Why you'll love working at PlayOn
Product, potential, and people. We're a leader in the high school event space, constantly evolving our product to meet the needs of administrators. We focus on solving real challenges, learning quickly, and creating impactful solutions.
This is a growth-stage company, meaning your contributions have real impact. You'll have opportunities to grow your skills, tackle meaningful problems, and make a difference in the lives of schools and the students and fans they serve.
Our culture is built on accountability, collaboration, growth, and fairness. We don't just show up-we show up for each other. Everyone wears the same jersey, and we play hard, make the extra pass, and cheer one another on. Losses teach us, challenges motivate us, and persistence drives us forward. We value integrity over shortcuts, choosing to do what's right even when it's hard. Together, we strive to be better every day-because we know that's how we win as a team.
The Benefits We Offer
Multiple medical insurance plans to choose from
Dental, vision life and disability insurance
Employee Emergency Fund
Company equity (stock options)
Open PTO policy
401K plan with company match
Hybrid/flexible work environment
Note: Must be a full-time employee to participate in the company's employee health benefit plan. Part-time employees and interns are not eligible to participate.
Compétences linguistiques
- English
Avis aux utilisateurs
Cette offre provient d’une plateforme partenaire de TieTalent. Cliquez sur « Postuler maintenant » pour soumettre votre candidature directement sur leur site.