Muhammad Usman Email & Phone Number

Lahore, Pk

1. Led a team of 25 data scientists in the fine-tuning of a large language model (LLM) using SFT and RLHF techniques, resulting in performance metrics crucial for real-world applications.2. Automated the detection of issues in SFT notebooks by leveraging LLM models, achieving a 20% improvement in accuracy and decreasing manual review time by over 30 hours weekly.

1. Incorporated BERT model for job-profile matching and implemented developer clustering, resulting in tailored treatments and boosted operational efficiency. Conducted funnel and survival analysis to define SLAs and optimize customer touchpoints.2. Designed and automated data pipelines with Apache Airflow, implementing task scheduling and monitoring to streamline data workflows, reducing report generation times by 35% and improving data accessibility. 3. Collaborated with… Show more 1. Incorporated BERT model for job-profile matching and implemented developer clustering, resulting in tailored treatments and boosted operational efficiency. Conducted funnel and survival analysis to define SLAs and optimize customer touchpoints.2. Designed and automated data pipelines with Apache Airflow, implementing task scheduling and monitoring to streamline data workflows, reducing report generation times by 35% and improving data accessibility. 3. Collaborated with cross-functional teams on feature requirements, conducted A/B testing, and reported insights, boosting product satisfaction by 25% and enhancing performance.4. Transferred key business intelligence dashboards from Mode to LookerML, enhancing data clarity and usability while reducing report generation time by over 30% for stakeholder teams relying on analytics. Show less

1. Led XGBoost-based propensity score classification, improving gain metric by 12%, and relayed actionable insights to cross-functional teams and management.2. Managed over $20M in client portfolios as Account Lead while implementing predictive models for effective customer-agent pairing; improved agent utilization rates by 25%, ensuring maximum operational efficiency.

Lahore, Punjab, Pakistan

1. Improved production monitoring by analyzing AI model performance and conducting root cause analysis, leading to faster issue resolution. Predicted caller routing patterns through exploratory data analysis, enabling more targeted service across demographics and regions.2. Developed interactive Power BI / Tableau dashboards using Dataverse data modeling, providing actionable insights to support data-driven decision-making.

Lahore, Pakistan

During my undergrad I had the chance to assist in teaching Signals and Systems and Calculus courses. My responsibilities included designing quizzes and leading tutorial sessions to enhance students' problem-solving abilities.

Laahore

Load forecast and Power planning