Omer Sajid

Software Engineer, Machine Learning @Yabla

New York, NY, US

Architected a continual learning video recommendation system leveraging hybrid collaborative filtering with two-tower neural architecture, integrating RoBERTa-based content embeddings and user language abilities.• Fine-tuned Qwen2.5-14B model with QLoRa using PEFT on 300+ curated multilingual interlinear gloss examples, increasing task accuracy by 23% and reducing educator workflow from 4 hours to 10 minutes per video.• Spearheaded full-stack development of \"Fluency Club\" engagement suite featuring daily challenges, community chat, tutor scheduling, personalized curriculum, progress tracking, and interactive speech recognition, increasing Monthly Active Users by 13% over 3 months.• Optimized video search infrastructure by migrating from SQL-based queries to an inverted-index engine, eliminating user complaints and reducing latency from 1.5 s to 150 ms for 800K+ daily requests.• Engineered cross-platform integration between sister companies using JWT-based server-to-server authentication and REST API gateways, enabling seamless interoperability.• Built end-to-end data pipelines encompassing both real-time streaming and batch ETL architectures with visualization dashboards, monitoring 12 KPIs across user engagement, retention and revenue metrics.

Lahore Grammar Senior School for Boys [LGS JT]

A level

DePauw University

Bachelor's degree, Computer Science

University of Illinois Urbana-Champaign

Master of Science - MS, Computer Science