Accenture
Data science and ML engineering in supply chain, operations, and GenAI. Delivered at scale across EMEA and the US.
Solution Migration to Private Cloud Environment
Delivered 17 end-to-end solution migrations in 4 months, aligning teams across EMEA and the US and ensuring seamless technical execution and client satisfaction.
- Daily coordination with cross-functional teams using Agile methodologies
- Creation of CI/CD pipelines to automate deployments and testing
- Planning and organisation of migration activities
- Deployment of AML endpoints on Azure Kubernetes Service for scalable model serving
- Weekly stakeholder meetings to ensure alignment and transparency
Python · Azure AML · Azure ADX · PowerBI · Azure Kubernetes Service · Azure Functions · CI/CD · Team Management
Optimisation of Chocolate Spraying Process in US factories
Delivered a scalable ML-driven solution to optimise chocolate spraying on peanuts, integrating a complex ETL pipeline and deploying computer vision models for real-time monitoring. Ensured seamless collaboration across teams and maintained strong stakeholder alignment throughout the project.
- Design and implementation of machine learning models to optimise chocolate spraying on peanuts
- Development of a complex ETL pipeline to process and integrate production data
- Deployment of computer vision models for real-time monitoring and quality assurance
- Creation of CI/CD pipelines to automate training, testing, and deployment
- Regular stakeholder sessions to ensure alignment with production teams
Python · Azure AML · Azure ADX · PowerBI · Azure Kubernetes Service · Azure Functions · CI/CD · Computer Vision · ETL
Forecasting and Mathematical Modelling for Detection, Prediction and Correction of issues in production plants
Delivered within a demanding 4-week RFP, incorporating forecasting models and mathematical modelling algorithms enhancing production to a 99.1% yield.
- Delivered within a demanding 4-week RFP, incorporating forecasting models and mathematical modelling algorithms enhancing production to a 99.1% yield.
Python · Databricks · Mathematical algorithms · PowerBI · PySpark
Algorithmic Solution for Material Traceability and Batch Analytics System
In this project, we primarily focused on developing an accurate mathematical model to identify efficiency bottlenecks linked to financial losses, enabling the site team to proactively address issues before incurring significant economic costs. Information was delivered to client via an intuitive collection of dashboards with deep analysis on their productions and schedules.
- Designed a track & tracing algorithm, and financial model achieving a 15% reduction in material overusage and improving financial assessment accuracy, deployed with CI/CD pipelines with 98.5% automated test coverage.
- Developed PowerBI dashboards, delivering KPIs, predictions and actionable insights for client decision-making, adopted by 6 production plants in the US.
Python · Azure Functions · Differential Algorithms · CI/CD · PowerBI · SQL · Azure DBs
High-Performance ML Model Enhancement and Code Optimization
Contributed to a mature project by boosting efficiency and model performance through strategic refactoring, documentation, and model retraining.
- Refactored codebase using hashmaps for efficient O(1) lookups.
- Retrained models with XGBoost to improve R² scores.
- Enhanced algorithm robustness with additional logic layers.
- Implemented a data quality assurance module with the team.
Python · Azure ML Studio · Azure DB · XGBoost · Azure Functions · ETL · Refactoring
ASGI-Based GenAI Platform for Seamless LLM Integration
Built a modular solution with a frontend, gateway, router, and SDK for GenAI agents, enabling seamless interaction with databases, audio, images, and documents, adding ML and CV capabilities.
- Implemented an ASGI architecture with FastAPI for GenAI with mathematical simulation, image-based fault monitoring & document processing, reducing plant production stoppages by up to 30%.
- Developed predictive maintenance ML models and integrated them with GenAI agents, achieving 81% accuracy in early failure detection and preventing potential downtimes.
- Deployed scalable microservices on Azure & Azure Machine Learning, supporting up to 3k+ concurrent users with Agentic AI deployments in MS Teams and Slack.
Python · FastAPI · Microsoft Bot Framework · Azure · ASGI · Pytest · Langchain · Streamlit