🤖 Lead AI Developer

Building Intelligent Systems That Scale

Specialized in Document AI, NLP, and LLM-powered applications. Professional results with measurable impact.

From auditing financial institutions to architecting AI systems—I've transformed my analytical foundation into cutting-edge machine learning solutions. Now leading AI development at a healthcare start-up, I turn chaotic medical documents into structured intelligence with zero false positives.

My mission: Build secure, private, and scalable AI systems that deliver measurable business impact. From robust ETL data pipelines to production-grade ML—every system I architect is built for enterprise, not just proof-of-concept.

4+ Years Exp.

50%→90% Accuracy Gains

~70% Speed Boost

Specialization
Document Intelligence & LLM Pipelines

Production Impact
83% Extraction Hit Rate / 5hr→30min Reports

Core Stack
AWS Bedrock, SageMaker, Neo4j, LangChain

Location
Hamilton, ON 🇨🇦

💼 Experience

Delivering measurable AI/ML impact across healthcare and finance

SiftMed

Lead AI Developer | NLP & LLM Expert

Jul 2025 – Feb 2026

▸ Improved page stream segmentation accuracy 50% → 90% using Strands Agents with foundational models (Claude Sonnet, Haiku) on AWS Bedrock

▸ Engineered intelligent redundant page detection system achieving zero false positives in duplicate identification—ensuring flawless document integrity using foundational models

▸ Established end-to-end ETL evaluation pipeline for production monitoring using AWS EventBridge, Step Functions, and MySQL

▸ Designing scalable document intelligence pipelines with AWS Bedrock, Lambda, Textract, Step Functions, and SageMaker

▸ Driving R&D initiatives for medical text extraction, summarization, and structured data generation

▸ Mentored junior engineers on LLM guardrails for hallucination prevention—identified key vulnerability points for prompt injection attacks and defined input/output validation rules with structural components to guide and restrict model behavior

▸ Guided junior developers in refining RAG pipelines—covering vector database setup, embedding strategies, and retrieval optimization for document-based Q&A systems

SiftMed

Senior Machine Learning Engineer | NLP & LLM Specialist

Oct 2022 – May 2025

▸ Accelerated processing speeds ~70% leveraging PySpark and Pandas for computational efficiency

▸ Fine-tuned state-of-the-art models (BERT, DiT, ViT, GLiNER) for token classification and NLP tasks using transfer learning

▸ Deployed LLMs/SLMs (Llama3, Gamma, Phi-3) for information retrieval and LLaVA1.6 for layout parsing

▸ Developed and fine-tuned ML/NLP models on AWS SageMaker for automated decision-making processes

▸ Built robust data pipelines on Amazon S3 for model training and inference workflows

▸ Contributed to infrastructure using Terraform, Docker, ECS, CloudFormation, and Step Functions

Royal Bank of Canada

Lead Data Engineer

May 2025 – Jul 2025

▸ Set up and maintained Neo4j graph database for enterprise knowledge networks

▸ Developed OCR APIs using PaddleOCR for large-scale document ingestion and processing

▸ Built secure ETL pipelines from NAS drives → Neo4j with validation layers for data consistency

▸ Automated GitHub-to-VM deployments via Ansible, improving delivery reliability

▸ Supported teams with Neo4j integration, API setup, and SageMaker environment provisioning

Nova Scotia Community College

Casual Research Associate

Sep 2023 – Feb 2024

▸ Achieved 83% hit rate on PIVP Invest Nova Scotia research project through model fine-tuning and RAG implementation

▸ Improved accuracy ~10% vs leading open-source models using fine-tuned datasets and LLM optimization

▸ Implemented RAG pipeline using LangChain, Pinecone, and fine-tuned 7B open-source model

▸ Overcame complex data extraction from web PDFs using LLMSherpa and LangChain for advanced chunking

▸ Delivered bi-weekly presentations to stakeholders, aligning with business objectives

CGI

Data Engineer – Internal Audit

Feb 2022 – Oct 2022

▸ Improved RPA BOT efficiency 80% by resolving bugs and creating features for business automation

▸ Built ETL pipelines using Dataiku for multi-million-dollar Risk Assessment Classifier project

▸ Developed anomaly detection reporting tool using Selenium for improved report accuracy

▸ Collaborated with data scientists and business stakeholders following Agile methodology

▸ Learned and adapted to Blue Prism and OpenShift for project requirements

⚡ Technical Skills

Comprehensive AI/ML technology stack

🤖 AI/ML & NLP

LLMs RAG Pipelines Prompt Engineering Document AI NLP BERT LayoutLMv3 GLiNER LLaVA Mistral Llama3

☁️ Cloud & MLOps

AWS Bedrock AWS SageMaker Docker Terraform Ansible CI/CD GitHub Actions

💾 Data Engineering

PySpark Pandas Neo4j Pinecone Vector DBs ETL Pipelines Dataiku

💻 Programming

Python (Expert) PyTorch Hugging Face LangChain FastAPI REST API

🚀 Featured Projects

AI/ML projects with real-world impact

📚

RAG Tutorial

Comprehensive Retrieval-Augmented Generation tutorial covering vector databases, embeddings, and LLM integration for intelligent document Q&A

LangChain Pinecone OpenAI Python

�

LLM Agent

Autonomous agents with tool-calling, memory management, and multi-step reasoning for document analysis and workflow automation

LangChain OpenAI API Vector DBs Python

🎯

Model Fine-tuning

Fine-tuning tutorials for GLiNER (NER), Kosmos-2 (vision-language), and Phi-2 (small language model) with practical examples

GLiNER Kosmos-2 Phi-2 HuggingFace

📄

Document AI

Fine-tuning and inference pipelines for document understanding models including LayoutLMv3 and UDOP for layout analysis

LayoutLMv3 UDOP PyTorch Transformers

🔗

GLinker Usecase

Practical examples demonstrating GLinker for entity linking and knowledge graph integration in NLP applications

GLinker NER Knowledge Graphs Python

🎓 Education

Academic foundation in computing and data analytics

M.Sc. Computing & Data Analytics

Saint Mary's University, Halifax, NS

2021 – 2022

GPA: 4.15/4.3 (Distinction)

B.Tech

Birla Vishvakarma Mahavidyalaya, India

2016 – 2020