Zaid Ahmad

Expert in Generative AI and Agentic AI, with hands-on experience developing and deploying advanced solutions for the U.S. mortgage industry. Delivered production-grade applications, including a ChatGPT-like assistant and RAG pipelines for intelligent document Q&A using mortgage forms (Form 1003, W-2s, paystubs). Leveraged Python, FastAPI, React.js, LangChain, Pinecone, and LandSmith to build scalable, real-time systems.

Key contributions:

• Designed and fine-tuned LLMs for document entity extraction and classification.
• Built RAG pipelines using LangChain and Pinecone to automate underwriting and loan review.
• Applied prompt engineering and agentic AI techniques for enhanced context-aware reasoning.
• Integrated OpenAI, Amazon Bedrock, and custom LLM APIs for real-time document intelligence.
• Extracted insights directly from Encompass, reducing manual effort, and improving transparency.

UI Development (React.js): Designed and developed dynamic chatbot UIs, connected to different knowledge bases. These UIs provided users with an interactive GenAI experience for document Q&A and loan guidance.
DevOps and Cloud Deployment:

• Containerized FastAPI apps using Docker, deployed on EC2 within VPCs for security and scalability.
• Integrated Redis for caching, Lambda for serverless workflows, and SNS for asynchronous notifications.
• Used API Gateway to manage scalable RESTful endpoints.
• Configured the Application Load Balancer (ALB) for routing and scaling.
• Implemented Amazon OpenSearch and Pinecone for indexing and querying document content within RAG pipelines.
• Managed deployment infrastructure with S3, CloudFront, and Route 53 for secure, global web app hosting.

• AWS: EC2, LB, VPC, Route 53, ECR, API Gateway, Lambda, S3, SQS, Elastic Redis cache, Bedrock, LLM (Cloud 3.5 Sonnet), OpenSearch.

• Azure: AI Search, Function App, OpenAI, LLM (GPT-4o, GPT-4o-mini), Foundry, Key Vault.

• Enhanced the performance of a CMS (Content Management System) developed in FastAPI and performed unit tests.
• Design a DAG in Airflow for various use cases.
• Created a Lambda function and wrote unit test cases using pytest, achieving 100 percent code coverage.
• Technology: FastAPI, AWS (RDS, Lambda, S3, SQS), Azure DevOps, Airflow, and MySQL.

• Design code that fetches live ECG data from AWS IoT Core, which communicates with the web using Socket.IO for web ECG plotting.
• Enhanced performance of the Rule Engine service and dashboard APIs by code and query optimizations, caching using Redis.
• Implemented Kafka-based communication between services, and significantly reduced errors.
• Technology: Flask, FastAPI, AWS (RDS, DynamoDB, S3), Redis, Kafka, Celery, PostgreSQL, Jenkins, and Grafana.

• Developed an API in Flask and used Celery to speed up performance.
• Worked on image processing, designed a machine learning model, and deployed it.
• Machine Learning - Developed model (Fertility Test) applied feature engineering, preprocessing data, trained the model and evaluate the model, and deployed it for inference.
• Technology: Django, Flask, Celery, AWS (RDS, S3, Lambda, DynamoDB), MySQL, and Shell.