Open to opportunities

Muhammad Sabih

@muhammadsabih1

Message

Senior Machine Learning Engineer building production LLM systems and scalable MLOps.

Pakistan

Message

What I'm looking for

I’m looking to build production-ready Generative AI and LLM systems end-to-end—strong MLOps, scalable pipelines, and observability—using RAG, human-in-the-loop validation, and multi-agent automation to improve performance and engineering productivity.

I’m a Machine Learning Engineer with 5+ years of experience designing and deploying production-ready AI systems. I specialize in Generative AI, LLMs, Multi-Agent Systems, and MLOps, with a proven track record delivering $3M+ annual GTV impact across search, recommendations, and trending algorithms.

At Careem, I architected and deployed an enterprise-grade LLM translation system using LangChain and FastAPI, eliminating external dependencies and achieving a 97% efficiency gain with a 10% accuracy improvement. I built an asynchronous translation pipeline with Langfuse for observability and Argilla for human-in-the-loop validation, optimized inference using RAG with Weaviate, and reduced outsourcing costs by 98% for legal document workflows.

I also focus on automating engineering workflows through multi-agent systems with LangGraph and MCP—developing an SDLC Agent for repository analysis, RFC/PRD generation, and effort estimation, plus an Engineering Intelligence Bot for real-time DevOps metrics. Previously at Confiz, I led ML initiatives that cut pipeline execution time from 5 days to under 12 hours and supported large-scale PySpark time-series model training while strengthening CI and data quality gates.

Experience

Work history, roles, and key accomplishments

Current

Software Engineer II - MLOps

Current

Careem

Jan 2024 - Present (2 years 5 months)

Architected and deployed an enterprise LLM translation system with LangChain and FastAPI, achieving a 97% efficiency gain and +10% accuracy; improved translation accuracy and efficiency using Langfuse observability and Argilla human-in-the-loop validation. Optimized RAG inference with Weaviate (+7% accuracy), built a document translation workflow reducing legal outsourcing costs by 98%, and develo

LLMs (LangChain fastAPI Retrieval Augmented Generation (RAG)Weaviate Langfuse Argilla LangGraph MCP Kubernetes

Senior Machine Learning Engineer

Confiz

Jan 2023 - Jan 2024 (1 year)

Designed a model optimization strategy using Python and Prophet, reducing pipeline execution time from 5 days to under 12 hours. Led an Agile ML team delivering markdown forecasting models, implemented Python backend with Dash/Plotly dashboards, and added quality gates that reduced error rates by 40%.

Python Prophet Quality Gates Plotly Agile Delivery Optimization

Machine Learning Engineer

Confiz

Sep 2021 - Dec 2022 (1 year 3 months)

Administered and fine-tuned large-scale PySpark pipelines using advanced hyperparameter techniques to train 1M+ time series models. Built model registry APIs to accelerate binning/registration/deployment, automated ML pipeline execution with CI practices, and improved ETL job runtime efficiency by 95%.

Pyspark Hyperparameter Tuning Time Series Modeling Model Registry APIs CI CD ETL Data Validation Version Control

Software Engineer

Confiz

Jul 2020 - Aug 2021 (1 year 1 month)

Migrated legacy systems to a scalable microservices architecture using Spring Boot, Kafka, Docker, and Azure Kubernetes Service, improving maintainability. Built and optimized high-throughput REST APIs (1M+ daily users, ~300ms response), improved observability with Prometheus/Grafana, and reduced service downtime by 30%.

Spring Boot Kafka Docker Azure Kubernetes Service REST APIs Prometheus Grafana Observability