Machine Learning Engineer specializing in production AI systems for enterprise automation. I build end-to-end ML pipelines with ASR, TTS, and multimodal LLMs that solve real-world problems at scale.

Machine Learning Research Engineer (Oct 2024 - Present)

Working on production AI systems including:

• Telecom Customer Support Automation: Built production voice AI pipeline (ASR → LLM intent router with function calling → RAG retrieval → TTS) automating telecom customer support with optimized low-latency inference, generating PKR 1.6 crore monthly revenue
• Insurtech QA Automation: Developed automated quality assurance system using audio multimodal LLM to analyze customer service calls and agent performance for waada.pk, reducing manual QA costs through automated audio transcription, sentiment analysis, compliance checking, and performance scoring at scale
• Audio Retrieval System: Built production audio retrieval pipeline with fingerprinting and query-by-humming capabilities for content identification and search, implementing reranking algorithms optimized for real-time audio recognition in high-throughput environments

Natural Language Processing Engineer Intern (Jun 2024 - Aug 2024)

• Fine-tuned Whisper Large-v3 for Roman Urdu transcription (Pakistan's first Roman Urdu STT), reducing WER from 100% to 33% through custom dataset curation and end-to-end pipeline optimization
• Architected low-latency ASR system by modifying Whisper transformer architecture for streaming inference

Programming: Python, C++, Java

ML/AI Frameworks: PyTorch, TensorFlow, Whisper, LLMs (local and third-party deployments), Transformers, HuggingFace, LangChain

Voice & Audio AI: ASR, TTS, Audio Engineering, Speech Processing

ML Engineering: Model fine-tuning, Inference optimization, Low-latency pipelines, On-device deployment, RAG systems, Function calling

Backend & Infrastructure: FastAPI, Flask, Real-time systems, Pipeline orchestration, WebSockets/streaming

DevOps & Deployment: Docker, Kubernetes, CI/CD, Cloud platforms (AWS/GCP/Azure), Model serving, Production monitoring

Databases: Vector databases (Pinecone/Weaviate/Chroma), SQL, NoSQL (MongoDB/Redis)

Bachelor's in Financial Technology
National University of Computing and Emerging Sciences (2023-2027)

A Levels
University of Cambridge (Grade: A* A B, 2021-2023)

• Email: masabnadir530@gmail.com
• LinkedIn: linkedin.com/in/m-masab
• Location: Islamabad, Pakistan