Wajid Karim

AI Engineer with hands-on production experience in multi-agent systems, LLM pipelines, and RAG architectures. Proven ability to build and deploy intelligent systems across healthcare, automation, and analytics domains.

Background in full ML pipelines from data annotation and model training to deployment and monitoring. Passionate about applying cutting-edge AI to solve real-world problems.

AI Engineer - Canvas Solution

(2026-02 - 2026-04)

• Built ASA-Agent, a production multi-agent healthcare assistant using LangGraph and LangChain with intent routing across scheduling, emergency, cancellation, and FAQ workflows.
• Designed dual RAG pipeline strategy (FaqNode & FaqNodeNoTool) with optimized chunking strategies, embedding models, and context window management for medical knowledge retrieval.
• Implemented semantic search via pgvector (PostgreSQL) for dense embedding storage and retrieval; applied re-ranking logic to improve answer faithfulness.
• Enforced structured outputs using Pydantic schemas; monitored model performance using MLflow to track inference quality and latency tradeoffs.
• Applied NLP techniques including intent classification and entity extraction to route patient queries across specialized agents.

AI Data Annotator / ML Intern - Unpeel (Startup)

(2025-12 - 2026-02)

• Collected and structured children's fashion product datasets across 19 brands for training ML classification models.
• Labeled and validated product attributes (gender, clothing type, brand) to support NLP-based text classification pipelines.
• Supported the full ML pipeline: data collection to labeling to preprocessing to model training preparation.

Machine Learning / Deep Learning Fellow - Buildables

(2025-07 - 2025-11)

Completed 3-month intensive fellowship in ML and Deep Learning; worked on research-driven practical AI projects.

• Received strong mentor recommendation for exceptional performance.

Junior Data Scientist & AI Engineer - IT Directorate

(2024-06 - 2024-12)

• Analyzed organizational data using Python and scikit-learn to build models supporting 3+ strategic decisions by department leadership.
• Built predictive dashboards for sales and customer analytics, reducing manual reporting time by ~40% and improving insight turnaround.
• Led AI-driven automation projects using TensorFlow/Keras, eliminating repetitive manual processes across 2 operational workflows.

Bachelor of Science - Computer Science - Karakoram International University (KIU)