Data & AI engineering

Muhammad Mudassir Khan

Independent data and AI consultant building production-ready systems with full-stack observability, safety guardrails, and evaluation. I lead and scale data teams and ship forecasting, MLOps, and agentic AI products across Q-Commerce, mobility, and enterprise.

Lahore, Pakistan · Open to relocation

Muhammad Mudassir Khan

AI-assisted development & AI agents

Today I focus on how large language models and agentic workflows change how we build and operate data and AI systems. I work with LangGraph-style orchestration, RAG, and evaluation loops (offline evals, RAGAS-style metrics, LangSmith tracing) so systems stay reliable in production—not just impressive in demos.

Recent work spans an AI agent–based RAG chatbot for internal policy search, EdTech assessment pipelines, and MLOps platforms with feature stores, drift detection, and CI/CD for training and deployment. I care about pairing AI-assisted coding with rigorous engineering: observability, guardrails, and clear success criteria.

Experience

Selected roles across leadership, delivery, and applied research.

  1. Independent Data & AI Consultant

    Self-employed · Oct 2025 – present · Lahore
    • Production data and AI systems with observability, safety guardrails, and evals.
    • AI-based EdTech assessment; MLOps for real-time fraud detection with feature store, drift detection, and CI/CD pipelines.

    GCP · LangGraph · LLMs · Postgres · FastAPI · Kafka · Flink · MLflow · BentoML · Airflow · Python

  2. Head of Data

    Waseela · Jan 2025 – Sep 2025 · Lahore
    • Technology leadership: org design, hiring, and career ladders for Data Engineering, Analytics, and Data Science (12+).
    • AI agent–based RAG chatbot for company policies; computer vision for animal ID on mobile.
    • Medallion architecture and automated data pipelines.

    AWS (Redshift, SageMaker) · Airflow · Airbyte · Metabase · Vector DBs · PyTorch · Python

  3. Manager Data Science

    Delivery Hero SE · Dec 2022 – Jan 2025 · Berlin
    • Led 8+ data scientists and engineers; delivery intelligence for Q-Commerce across 50+ countries.
    • Demand forecasting and planning platform replacing legacy tooling; improved availability by up to ~5% while controlling stock-on-hand.
    • Promotion elasticity modeling, picking strategy optimization, and scalable ELT with observability and SLAs.

    GCP (BigQuery, Cloud Functions, Cloud Run) · Airflow · Python · SQL · Looker

  4. Director of Data & BI

    Retailo Tech · Sep 2021 – Nov 2022 · Lahore
    • Built and scaled a 15+ person org (engineering, analytics, DS, annotation) for marketplace growth.
    • Data-as-a-product: procurement engine, search & recommendations; batch and streaming pipelines.

    AWS (S3, SageMaker, Redshift, QuickSight) · dbt · Kafka · Flink · Argo · TensorFlow · Python

  5. Technical Lead Manager

    Motive (formerly KeepTruckin) · Oct 2017 – Sep 2021 · USA (remote)
    • AI dashcam safety event detection for 150k+ vehicles; strong lift vs. industry benchmarks.
    • Sensor fusion and analytics for accident detection; GPS-scale stop classification.

    AWS (S3, Lambda, Redshift, IoT) · Kafka · TensorFlow · Python

Open source

Recent public repositories on GitHub.

ibook_mlops

AI Ops system deployable locally or in production, with a simulation module for scenario testing.

Python · Docker

iBud

Agentic RAG for e-commerce customer support: FastAPI backend, workflow-style agent runtime, pgvector/Redis, Streamlit UI, observability with Prometheus, Grafana, LangSmith, and RAGAS.

Python

Teaching & resources

Introduction to Machine Learning — course materials hosted on this site.

Writing

Selected articles on Medium — MLOps series and earlier work from Motive (KeepTruckin).

More on Medium (@mmk.wazir)

Research & education

Background

Doctoral research in robotics and computer vision at LUMS (not completed), including visiting research at RPTU Kaiserslautern-Landau, Germany. Thesis topic: safe roadmaps and traversability analysis using 3D sensors for autonomous vehicles. Member of the Cyber Physical Networks (CYPHYNETS) research group supervised by Dr. Abubakr Muhammad.

That work connects graph-theoretic road analysis with vehicle models to assess whether roads are traversable for a given vehicle in the presence of obstacles and how safety compares to nominal conditions.

Degrees

  • Doctoral research — Robotics & Computer Vision, LUMS (Jun 2012 – Oct 2017, not completed)
  • M.S. Computer Engineering — LUMS (Aug 2010 – Jun 2012)
  • B.E. Computer Systems Engineering — NUST (Sep 2004 – Jun 2008)

Publications

  • H. Ali, A. Muhammad, M. M. Khan. “A Simple Framework for Context-Aware Driver Performance.” IEEE ITSC, 2020.
  • M. M. Khan, H. Ali, K. Berns, A. Muhammad. “Road Traversability Analysis Using Network Properties of Roadmaps.” IEEE/RSJ IROS, 2016.

Skills

Data engineering

PythonSQLETL/ELTBatch & streamingOrchestrationData modelingGovernanceObservability

Cloud

GCP (BigQuery, Cloud Run, Vertex AI)AWS (S3, Redshift, SageMaker, Lambda)

MLOps & platforms

MLflowBentoMLAirflowdbtKafkaFlinkFeature storeModel registry

Data science & AI

Statistical modelingCausal inferenceMLDeep learningPyTorchAI agentsRAG

Languages: English (fluent), Urdu (native), Pashto (native).

Contact

mmk.wazir@gmail.com · Phone: +92 345 979 8131

LinkedIn · GitHub