Kunal J

Data Scientist

Best AI Sports Platform | AI for Sports Highlights & Commentary • Tiempo completo

Mar 2025 - Present • 1 yr 7 mos

I build the computer vision systems behind a video intelligence product. Day to day that means: - A multimodal pipeline combining object detection (YOLO), OCR and audio analysis for real-time event detection in live video, which cut manual tagging effort by around 70% - A replay detection system running at 99% accuracy in production, replacing a manual review process - A 16:9 to 9:16 video reframing pipeline with contextual subject tracking, fast enough to run in real time for automated short-form content The work is production-first: hard latency budgets, monitoring, and models that keep working after launch rather than demos that only hold up on the sample clip.

Data Analyst

Limeroad • Tiempo completo

Nov 2024 - Mar 2025 • 4 mos

Analytics and data modelling for a large e-commerce catalogue. - Found and fixed a broken base data table that had been silently corrupting reporting downstream, restoring integrity across every dependent dataset - Built the core reporting tables and standardised KPI definitions so dashboards across teams finally agreed with each other - Documented the data logic so the work stayed reusable instead of living in one person's head Useful background for any project where the data has to be trusted before a model goes anywhere near it.

Full Stack Developer

Tongadive • Tiempo completo

Jul 2023 - Mar 2024 • 8 mos

Backend and full-stack development on production systems. - Optimised backend data access and query performance, improving analytical response times by 40% - Implemented secure coding practices and AWS Well-Architected Framework principles across multiple certification cycles - Debugging, root-cause analysis and performance tuning on live systems This is the reason I'm comfortable owning deployment and APIs, not just the model.