- Led a team of 4+ AI/ML developers, driving a 35% increase in model efficiency and production.Implemented TensorFlow and PyTorch across 4+ projects, reducing computational costs by 25%.- Deployed AI models on AWS, improving scalability and supporting 15% more concurrent users.- Integrated advanced ML techniques to achieve a 50% improvement in predictive accuracy.- Upgraded AI infrastructure, enabling processing of 2TB+ datasets, a 60% increase over previous capabilities.- Designed and implemented an AI-powered assistant using the LLaMA2 7B model, fine-tuned with custom data, reducing support and data analyst workload by 45%.- Leveraged Hugging Face API for model weights, applied quantization and LORA training, and integrated Retrieval-Augmented Generation (RAG) for enhanced performance.- Hosted on a Linux machine with VLLM for optimal efficiency.- Delivered a robust AWS-based CISO bot to manage routine inquiries, reducing response times by 35%.- Trained to validate customer files against compliance checklists, improving productivity.- Integrated with SLACK API for seamless communication.- Developed on AWS using Claude-3 Sonnet LLM, with advanced prompt engineering and metadata enhancements. Utilized Code-Catalyst for CI/CD, ECR, and ECS for deployment.-Engineered comprehensive data pipelines to synchronize data from multiple systems into a centralized data lake for actionable reporting.- Utilized DataStage, Python, PySpark, and shell scripting for data transformations, with SQL-driven reporting presented in Excel.

- Developed an innovative recommendation system that automated the process of matching names between two different systems, cutting manual work by 50% and improving accuracy.- Used BERT to generate embeddings and applied cosine similarity to create highly effective recommendations. Despite its simplicity, this solution had a substantial impact on business efficiency.- Created a multilingual translation solution that integrated directly with existing ETL workflows, moving from cloud-based translation to an on-premises setup, which reduced cloud costs by 30%.- Led the entire project, from development to delivery, ensuring smooth coordination with stakeholders.- Built using PyTorch, the system included custom neural network models to handle multiple languages. - Engineered seamless ETL workflows to bring data from multiple systems together, enabling the generation of comprehensive reports on a weekly, monthly, and ad-hoc basis.- By identifying key attributes, I made it easier to map data across systems and streamline analysis, ultimately consolidating everything into a single source of truth. This allowed us to decommission outdated projects and make data access far more efficient.- Regularly tasked with producing detailed reports and managing cross-system data mappings using tools like DataStage, SQL, PL-SQL, views, procedures, and triggers.

- Implemented data validation processes to ensure the accuracy and completeness of transferred data from T24 to the DB2 data warehouse.- Collaborated with the IT team to optimize the performance of the DB2 data warehouse, improving query response times and overall system efficiency.- Conducted thorough testing and validation of the migrated data to guarantee data integrity and consistency post-migration from SQL Server to DB2.- Developed automated monitoring and alerting systems to proactively identify and address any potential data integration or migration issues.- Provided training and documentation to the bank's staff on the new database infrastructure and reporting tools to ensure smooth adoption and utilization of the updated systems.