A proficient AI/ML developer specializing in building scalable data-driven solutions,
advanced NLP pipelines, and innovative AI-powered applications.
Skills
Python
JavaScript
C++
MongoDB
AWS
Power BI
Flask
Machine Learning
FastAPI
NLP
LLMs
RAG
Experience
AI/ML Developer - Sheridan Centre for Applied AI
Sep 2024 – Present | Oakville, ON
Developed an automated pipeline leveraging Meta Llama 3 LLM to extract key fleet management KPIs from unstructured data sources.
Enhanced KPI extraction accuracy by over 20% through strategic prompt engineering and few-shot learning techniques.
Collaborated with industry partners to define requirements and deliver AI solutions aligned with business needs.
Machine Learning Data Scientist - Sheridan College
May 2024 – Aug 2024 | Oakville, ON
Engineered Python scripts for comprehensive data exploration and preprocessing of 30,000+ osteoporosis patient records.
Improved predictive model accuracy for fracture risk assessment by 40% by implementing an ensemble voting technique combining multiple ML models.
Performed feature engineering and selection to identify key predictors of patient outcomes.
Projects
Advanced RAG-Based Chatbot
Built a document-aware chatbot leveraging a Retrieval-Augmented Generation (RAG) architecture with AWS Bedrock (Claude v2 LLM) for accurate, context-specific responses based on private documents.
Developed an automated document processing pipeline using AWS Lambda and S3 triggers to generate vector embeddings (via amazon.titan-embed-text-v1) stored in OpenSearch.
Designed and deployed a serverless, user-friendly web interface using AWS Amplify, ensuring seamless interaction and low-latency query handling.
Integrated OpenSearch for efficient vector similarity search and CloudWatch for comprehensive monitoring, logging, and debugging.
SnapCal - Food Image Recognition App
Developed a web application using Flask that allows users to upload food images and receive nutritional information estimates.
Integrated Google's pre-trained Inception v3 model for image classification to identify food items.
Implemented logic to map identified food items to a nutritional database (e.g., USDA FoodData Central API or a local database) to retrieve calorie and macronutrient data.
Designed a simple user interface for image upload and results display.
