Israr A.
Senior Mechanical Engineering AI Evaluator
Experience
Jul 2019 - Present
6 years 5 monthsUnited States
Senior Mechanical Engineering AI Evaluator
JLL
- Led AI evaluation for mechanical engineering model validation using Python, NumPy, Pandas, and SciPy to preprocess and analyze simulation datasets from ANSYS, COMSOL Multiphysics, and ABAQUS.
- Designed and executed model evaluation pipelines in TensorFlow, PyTorch, and exported inference to ONNX for cross-platform validation and benchmarking with TensorBoard.
- Implemented model training reproducibility and experiment tracking using MLflow and Weights & Biases, integrating metrics into Prometheus and dashboards in Grafana for SLA monitoring.
- Built containerized evaluation environments with Docker and orchestrated batch evaluation and A/B testing using Kubernetes and Azure DevOps pipelines.
- Automated data validation and schema checks for simulation outputs using Great Expectations, Parquet data lakes, and Apache Arrow for high-performance I/O.
- Collaborated with simulation teams to convert ANSYS, LS-DYNA, and OpenFOAM output into structured HDF5/Parquet datasets, analyzed with Pandas and visualized in ParaView and Matplotlib.
- Applied explainability techniques SHAP and LIME on ML surrogates trained on MATLAB/Simulink and SolidWorks generated features to root-cause model drift and bias.
- Developed performance test suites with pytest, pytest-benchmark, and integrated with GitHub Actions and CircleCI for CI/CD gating of model changes.
- Led cross-functional reviews with data scientists and mechanical engineers to align AI outputs with engineering tolerances from COMSOL and ABAQUS scenarios.
- Implemented inference hosting and monitoring using AWS SageMaker and Google Cloud AI Platform, instrumenting logging via Prometheus and traces via gRPC endpoints.
- Designed RESTful APIs with Flask/FastAPI and gRPC endpoints to serve validated model predictions to engineering tools and dashboards.
- Performed statistical validation and uncertainty quantification using scikit-learn, SciPy, and custom Monte Carlo pipelines to evaluate model reliability under variable loads.
- Optimized data pipelines for large simulation archives using PostgreSQL for metadata, MongoDB for unstructured logs, and CSV/Parquet storage for bulk artifacts.
- Created model optimization recommendations: pruning with TensorFlow Model Optimization, quantization-aware training, and conversion to ONNX to improve latency for engineering inference.
- Conducted fault-injection and robustness tests using synthetic perturbations derived from SolidWorks parametric studies and OpenFOAM CFD variation sets.
- Documented evaluation procedures, validation criteria, and acceptance tests in versioned artifacts stored in Git, with release automation through Azure DevOps.
- Mentored junior engineers on integrating MATLAB simulation outputs into ML training sets, and best practices for reproducible experiments with Jupyter notebooks and Weights & Biases.
- Spearheaded initiatives to standardize model evaluation metrics (MAE, RMSE, R2, calibration, reliability diagrams) across teams and instrumented dashboards to track these KPIs using Grafana.
May 2014 - Jun 2019
5 years 2 monthsUnited States
Software Engineer
RealPage, Inc
- Developed ML pipelines for engineering data analysis using Python, Pandas, NumPy, and scikit-learn to validate surrogate models for mechanical behavior.
- Integrated simulation outputs from ANSYS and ABAQUS into training datasets, using ParaView and Matplotlib for visualization and inspection of potential failure modes.
- Built prototype neural surrogate models in Keras and TensorFlow, performed cross-validation, and exported models to ONNX for interoperability.
- Implemented experiment tracking and model versioning with MLflow and used TensorBoard for training diagnostics and hyperparameter tuning.
- Authored automated test suites with pytest and benchmarked inference performance across CPU/GPU using pytest-benchmark and Docker containers.
- Collaborated with DevOps to deploy model evaluation workloads on AWS SageMaker and container registries; used Git and GitHub Actions for CI.
- Performed statistical analyses and uncertainty quantification using SciPy and Monte Carlo sampling to quantify model confidence for engineering use cases.
- Used MATLAB and Simulink to generate labeled datasets and to verify ML model outputs against physics-based simulations.
- Applied explainability tools (SHAP, LIME) to interpret surrogate predictions and guided feature engineering from CAD exports (SolidWorks).
- Documented findings, created reproducible Jupyter notebooks, and delivered technical reports outlining model limitations and recommended remediation strategies.
Jul 2012 - Apr 2014
1 year 10 monthsUnited States
Software Engineer
Foreflight
- Contributed to engineering analytics and early ML model evaluation using Python, NumPy, Pandas, and scikit-learn to analyze flight and mechanical telemetry.
- Processed time-series and sensor data using SciPy and custom signal processing pipelines, visualized results with Matplotlib and ParaView.
- Built and validated prototype regression models in scikit-learn and Keras, assessing performance with cross-validation and statistical metrics (RMSE, MAE).
- Worked with CAD-derived inputs from SolidWorks for feature extraction and prepared simulation-driven training sets from ANSYS exports.
- Packaged evaluation scripts into Docker containers and maintained code in Git repositories integrated with CircleCI for automated tests.
- Developed basic REST APIs using Flask to serve model predictions to internal tools and dashboards.
- Implemented monitoring of model performance baselines and drift detection using custom scripts and integrated logging with MongoDB.
- performed sensitivity analysis and scenario testing using Monte Carlo methods and SciPy optimizers to validate model robustness.
- Collaborated with cross-disciplinary teams to align ML model outputs with mechanical engineering acceptance criteria and produced comprehensive validation reports.
- Maintained Jupyter notebooks and internal documentation demonstrating reproducible evaluation workflows and experiment artifacts tracked with MLflow.
Summary
Seasoned engineering leader with 13+ years driving AI-augmented mechanical engineering validation and model evaluation.
Proven track record delivering rigorous model assessments, reliability testing, and domain-aligned optimizations that reduced model error and accelerated deployment timelines.
Hands-on with industry toolchains for simulation, ML evaluation, and MLOps—bridging mechanical engineering and data science to produce production-ready AI systems.
Languages
English
NativeEducation
University of Houston-Clear Lake
Bachelor of Science · Computer Science · Houston, United States
Certifications & licenses
AWS Certified: Machine Learning Foundations
Certified SAFe® 5 Agilist
Microsoft Certified: Azure Developer Associate
Need a freelancer? Find your match in seconds.
Try FRATCH GPT More actions
Similar Freelancers
Discover other experts with similar qualifications and experience
- English
- Deutsch
2025 © FRATCH.IO GmbH. All rights reserved.