على عباس

▸ Built and deployed production AI backends using Python, Flask, and FastAPI with real-time
LLM inference
▸ Integrated OpenAI APIs and Hugging Face models to create intelligent automation and
content processing pipelines
▸ Designed and maintained RESTful APIs for AI model serving, achieving low-latency
responses under high load
▸ Implemented RAG (Retrieval-Augmented Generation) pipelines with vector databases
(FAISS, Pinecone) for enterprise knowledge platforms
▸ Optimized GPU-accelerated inference using CUDA-enabled PyTorch, improving throughput
via batching and async request handling
▸ Containerized AI services with Docker and contributed to CI/CD pipelines for automated
testing and deployment
▸ Led fine-tuning of mBART transformer models for Urdu ↔ Roman Urdu translation,
handling large-scale dataset preprocessing
▸ Developed distributed ML recommendation systems using PySpark and XGBoost for ~1M
user workloads

▸ Built Python/Flask backend services and integrated external APIs with database-driven
components
▸ Followed SDLC best practices including code reviews, testing, and version control
workflows