Mohammed Nowfal

Dynamic MLOps Engineer and Data Scientist with hands-on experience managing production-grade ML infrastructure and deploying scalable data pipelines. Built and Deployed Machine Learning and Artificial Intelligent Models to production. Proven track record in maintaining Airflow on Kubernetes, optimizing CI/CD pipelines, and automating containerized deployments using Terraform, Helm, and Docker.

Adept at bridging the gap between data science and platform engineering by building and deploying ML models, recommendation engines and implementing APIs that drive measurable business impact.

Data Scientist at Saks Global India (2024-01 – Present)

• Infrastructure & CI/CD Pipelines. Orchestrated cloud-native ML infrastructure on AWS EKS with secure secret management and custom Docker images, accelerated deployment by slicing CI/CD build time by 20%, optimized data jobs for Spark pipeline runtimes by 50%.
• Content Personalisation Engine, Built on embeddings and FAISS vector search to optimise editorial email assets and homepage banners, improved engagement and content relevance at scale. 30% uplift in relevant content sends.
• Designed and deployed a Browse Abandonment ML ranking model for email marketing, enabling trigger-based outreach to high-intent users and achieving 92% accuracy and recall, driving improved conversion performance.
• Cross-Retailer Product Scraping System, Architected a scheduled, containerized web scraping and processing pipeline via Selenium and Docker to scrape and ingest product data into Snowflake. Enabled config driven job submits, to allow setup of any website
• Multi-Modal AI Matching, product matching system leveraging weighted image/text embeddings and cosine similarity to achieve 92% accuracy at a 0.9 threshold, integrated Claude 3.5 - validate borderline matches (threshold based).
• Model governance & evaluation tool, reduced model evaluation and A/B testing time by 85%. Brought down time from 1 week to less than 1 day, with zero manual effort.
• Agentic Purchase-Order Pipeline, reduced PO prep from 5 min to 10 sec with zero errors. Freed 80 hrs/week for the 10-person team.
• Creative QA & Merchandising agentic tool delivering 20x reduction in manual QA effort, saving 20–25 hrs/week across a (3-person team). Validate product placement across arrays and filters at 100% accuracy, replacing 10x of manual effort and saving 40 hrs/week.
• Implemented a Rust API to serve brand-level best-seller rankings, including migrating sales data from Snowflake to SurrealDB to support low-latency query workloads.
• Engineered a daily email feature library pipeline, generating user-level engagement features from send, open, and click events across multiple time windows for analytics and ML consumption, a similar project for store feature library
• Led a security-focused migration of Snowflake authentication across Python, dbt, and Spark workloads, transitioning from password-based access to key-pair authentication to improve security, compliance, and consistency, with 0 downtime.
• Analyzed product returns data to identify key trends and drivers. Additionally trained a machine learning model to predict product return probability. Built a pipeline for validation that saved 2 hours per iteration.

Bachelor of Technology in Electrical and Electronics Engineering – University of Visvesvaraya College of Engineering, Bangalore (2020-12 – 2024-06)