AI-Powered Personalization Engine for Fashion & Retail
Aam Consultan
- Designed and implemented a multimodal, session-based recommendation system combining image and text data to enhance product discovery and personalization.
- Built visual and text-based product search using OpenAI and recommendation pipelines powered by the NVIDIA Merlin ecosystem.
- Automated product tagging using deep learning-based image classifiers for eastern and western wear, improving inventory accuracy.
- Contributed to Agentic, a conversational AI interface offering personalized recommendations and real-time customer support.
- Developed an AI-powered merchandising dashboard with dynamic rules and filters driven by customer behavior and business logic.
Avanceon – Lucky Core Industries
- Developed a machine learning solution to predict fiber quality KPIs at intermediate stages of production.
- Reduced dependency on end-stage quality checks by enabling earlier quality assessment during manufacturing.
- Improved operational decision-making and reduced rework through real-time quality predictions.
Avanceon – Pakistani Tobacco Company
- Built autoregressive models to forecast tobacco waste during preprocessing at the Akora Khattak facility.
- Identified patterns in historical production data to optimize tobacco processing.
- Delivered actionable insights to reduce material waste and improve processing efficiency.
Avanceon – Mari Petroleum Company Limited
- Optimized gas plant pump operations to enhance efficiency and reduce energy consumption.
- Developed predictive maintenance models for gas whistles and pipelines to detect anomalies and prevent failures.
- Improved reliability of gas infrastructure and minimized unplanned downtime.
Avanceon – Tapal
- Analyzed production and sensor data to monitor and optimize quality KPIs during manufacturing.
- Applied machine learning models to predict product defects and production bottlenecks.
- Built algorithms to forecast breakdowns and quality issues, improving process control and product consistency.
- Collaborated with operations to implement predictive insights, reducing downtime and increasing OEE.
Avanceon – DhaHubCo
- Optimized industrial chiller operations by analyzing temperature, load, and energy data in real time.
- Used regression-based models to determine efficient load distribution and runtime schedules.
- Implemented predictive control logic to minimize excessive cooling and reduce electricity costs.
- Integrated models into daily plant operations via dashboard monitoring tools.
Avanceon – Shell
- Led a project to forecast petroleum fuel shrinkage during transportation using SARIMAX and autoregressive models.
- Collected and analyzed GPS, topographic, weather, and route data to detect high-shrinkage patterns.
- Integrated weather APIs to correlate environmental impact on shrinkage.
- Recommended routing and handling strategies to reduce fuel loss during transit.
Odyssey Analytics
- Developed automated workflows using Alteryx for financial data transformation, integration, and reporting.
- Improved processing speed, accuracy, and visibility into financial analytics for decision-making.
Odyssey Analytics – Texas, USA
- Performed time series forecasting using SARIMAX, LSTM, and hybrid models on crypto mining data.
- Addressed missing values through imputation strategies (forward, backward, mean) to ensure dataset integrity.
- Predicted 7-day trends to optimize energy usage and mining strategy.
Odyssey Analytics – Texas Energy Company
- Built a centralized data lake to unify operational datasets across power plants.
- Designed ETL pipelines using AWS Redshift, Oracle, Kinesis, Lambda, and JAMS Server.
- Enabled scalable data access for analytics and business intelligence.
- Applied unsupervised learning and transformer-based models (DeepSCAN) to analyze genomic data.
- Built clustering models to segment patients by gene expression for precision medicine.
- Predicted treatment responses using genomic patterns for personalized healthcare solutions.