NOUMAN IRSHAD

AI Engineer

IBM • Full time

Jul 2025 - Mar 2026 • 8 mos

Worked as an AI Engineer focused on building production-ready Generative AI systems, RAG pipelines, AI automation tools, and scalable backend architectures for enterprise and research-driven applications. Core Expertise Designed and developed advanced RAG (Retrieval-Augmented Generation) systems using LangChain, LangGraph, LlamaIndex, and CrewAI Built intelligent multi-agent workflows with hybrid retrieval architectures combining BM25 + vector search Engineered scalable AI APIs and backend services using FastAPI, Flask, Docker, and RESTful architectures Worked with vector databases including Qdrant, Pinecone, ChromaDB, and FAISS for semantic search and knowledge retrieval systems Developed enterprise-grade AI automation systems with clean modular Python architecture, testing workflows, Git version control, and SQL/NoSQL integrations Generative AI & LLM Engineering Worked extensively with modern LLMs including GPT-4, Claude, Gemini, Llama 2, BERT, and T5 Applied advanced prompt engineering, context optimization, and retrieval orchestration techniques for high-accuracy AI systems Implemented model fine-tuning approaches using LoRA/PEFT for domain-specific optimization and efficient inference workflows Built intelligent conversational systems, AI copilots, and document-aware assistants using LlamaIndex and LangGraph workflows Machine Learning & Deep Learning Experienced with supervised and unsupervised learning pipelines, feature engineering, preprocessing, and model optimization Worked with deep learning architectures including LSTM, RNN, and ANN Utilized frameworks such as PyTorch, TensorFlow, Keras, Scikit-learn, NumPy, and Pandas for ML experimentation and deployment Key Contributions Developed AI-driven automation tools including a Prompt-to-JSON Enhancer and AI Content-to-PDF Enhancer using FastAPI with Gemini/OpenAI APIs Fine-tuned Facebook’s MMS/TTS speech model for multilingual speech synthesis, evaluated architectures including XTTSv2.

Junior AI Engineer

Alibaba Group Holding • Full time

Jan 2024 - Feb 2025 • 1 yr 1 mo

Built end-to-end RAG systems using LangChain + Pinecone/ChromaDB for enterprise information retrieval, including hybrid search pipelines — applicable to intelligent document processing. Developed GPT-3/4-powered agentic workflows using LangChain and LangGraph for multi-step reasoning; deployed via clean REST APIs for cross-functional engineering integration. Designed a RAG-based search system with custom embeddings in a microservices architecture — full model training, evaluation, and bias reduction cycles. LSTM + clustering hybrid model on Parquet/CSV financial datasets — 22% improvement over ARIMA baseline; NumPy/Pandas feature engineering with statistical preprocessing, directly applicable to quantitative finance use cases. Engineered a GPT-powered AI chatbot ( +50% client engagement ); custom text embedding models integrated with vector databases for semantic search over unstructured data. Developed custom text embedding models and integrated them with vector databases to power semantic search and retrieval capabilities for unstructured text data.