Ali B. Email & Phone Number

Dearborn, Michigan, United States

Agile AUTOSAR/FNV4 Software Developer:*SAFe Agile certified; transitioned team to agile process.*Transitioned ASIL rated SW Components to AUTOSAR compliant models utilizing both bottom-up & top-down approaches.*Developed embedded safety ISO 262626 compliant SW in C for all xEV powertrains under Agile.*Authored AUTOSAR TSR/SSRs & linkages in Jama & Magic Draw; utilizing P/Behavioral/Activity/Fishbone Diagrams. *Identified AUTOSAR IO interfaces in PREEvision & completed ARXML handshaking.*Transitioned releases SW Component release to GitHub from Clearcase.Motor Generator Independent Plausibility (MGIPC) Check SW Pre-FNV4 feature owner*Responsible for MGIPC C SW development & feature bookshelf for all xEV programs.*Managed official and experimental labels in ClearCase & generated application builds.*Created URDs, dispositioned issues in CCB/JIRA forum identifying PCA/ICA resolution for launch issues.*Utilized PolySpace, Smartbear, BeyondCompare, to ensure quality MISRA C SW.Software Validation & Unit Testing*Created SW component concept models in Matlab/Simulink/C & developed test harnesses.*Unit Tested C++ SW utilizing BTC, Ceedling, Unity, Simulink test achieving MCDC coverage. Enforced MISRA standards.*Leveraged HIL, and breadboard to validate SW.*Performed vehicle testing utilizing ATI, no-hooks, Kavaser & myCANIC to validate SW.Owner of DVM (Design Verification Method) for One Pedal Drive, Driver Demand, & Low Mu testing.*Developed DVMs from scratch.*Championed data analysis for xEV program sign off & generated V&V report.*Analyzed data in ATI & SW implementation; used RCA to identify issues & write JIRAs against them.*Reviewed reqs. in Magic Draw & led effort to develop with the help of HARA/FMEAs.Big Data Utilization*Developed scripts using Julia/Python in Jupyter to parse event based data from customers & generate reports of DTC occurrences & status for relevant modules to get ahead of high-occurrence DTCs before vehicle launch.

•Identified downsides to current target-based Camera calibration method•Researched multiple target-less methods that would better meet customer expectation & reduce need for calibration booths, tooling, & work hours •Utilized Domino, Linux, Docker to create flashable environments & work with large LIDAR/Image data as well as Structure for Motion to yield workable Camera point cloud•Implemented & Presented algorithm using C++, ROS, & python to automatically identify discontinuous edges in LIDAR & Camera point clouds registering them together correcting for errors & more accurately being able to detect vehicle surroundings•Tracked progress using Gantt chart•Delivered proof of concept to DAT team for further development

Applied KnowledgeLanguage: Matlab* Applied LQR trajectory tracking on Suspension Model and Bicycle model.* Utilized quadratic programming MPC control to design a controller that follows a given trajectory based on the kinematic bicycle model* Utilized ICP tp find rigid body transformation to transform source to target point cloud* Computed discrepancy map in a set of stereo images and used SURF for feature extraction then generated codeworks using k-means clustering for the image set. This was used to match a query image to the closest histogram BoW in Euclidian distance.* Designed a PID controller to generate a trajectory for a non linear bicycle model to traverse a given track with left and right borders in a given time.

Participate in multiple case studies using CarSim to explore effects of road loads, ride, cornering, steering, suspension, and tires on vehicle performance and driver motion sickness under low/split/high mu conditions. Utilized power spectral density, jerk, and lap time to evaluate performance and effects on motion sickness.

*Paper Flattening and Template Matching Group Project:Language: Python - Generated a set of synthetically crumbled papers, their 2D flow graph, and mask images for our dataset - Trained a U-net to take in a wide variety of patches from the distorted images and output their respective flow maps (mapping each pixel of the distorted image to the original image) - Utilized template matching and K-means clustering to detect the numbers and operation in each math problem for the flattened math worksheets and generate the answer key https://github.com/abadredd/the-flatteners*Applied Knowledge:Language: Python - Estimated homography between two NFL pitch images to generate yellow line for first down marker - Applied demosaiking on image using Bayes filter to generate colored and grayscale images and reconstruct image - Utilized Harris corner detector and gaussian blurring to identify relevant features/descriptors to estimate a homographay for image stitching - Implemented a DoG blob detection and scale selection algorithm to detect disks of sunflowers in image- Implemented foreground-background graph cut using super pixels to generate an adjacency matrix and a capacity matrix for a key feature in the foreground based on histogram of intensities and spatial proximity. Then used Ford Fulkerson to find the maximum flow through the graph to perform the graph cut.- Implemented a CNN and a Multi-layer perceptron NN to classify images in the CIFAR-10 dataset

*Group Project: Semantic LIDAR OdometryLanguage Used: C++, Matlab -Utilized semantic KITTI dataset and applied SICP (semantic-ICP) to generate pose transformations between consecutive point clouds. Fed these transformations as edges into ISAM to generate smoothed trajectory -Compared odometry results between smoothed vs unsmoothed trajectory, and GICP vs SICP against group truth -Explored effects of increasing time step between point clouds used to generate transformations and removing semantic labels for moving objects on odometry results -Attempted SLAM by generating loop closure transformations when the vehicle crossed (came close to previous poses) and fed these loop closures as part of the factor graph to ISAM2 https://github.com/tglenn28/SLO https://www.youtube.com/watch?v=0wvZ5xyvVrM&t=2s*Designed Kalman Filter in V2I ImplementationLanguage Used: Matlab -Simulated a V2I environment by generating noisy measurements from 3 separate radar towers that will return the distance of the vehicle from each radar tower at each time step. -Propagated noisy measurements through a KF to estimate the current state of the vehicle in the 2D environment. -Compared the results to the ground truth assuming no knowledge of the vehicle motion (random walk motion model) starting with an initial guess for the state. https://github.com/abadredd/Exploring-Kalman-Filter-in-a-V2I-Implementation.git*EKF, UKF, PF, and in EKF Implementation using Velocity Motion Model in SE(2)Language Used: Matlab -Derived propagation and correction equations for each filter -Implemented filters in Matlab (prediction and correction) on set of noisy measurements received by simulated landmarks vs sensor readings for robot range and bearing measurements to generate odometry -Compared filter results

Language used: JuliaIDE used: Jupyter NotebookDocumentation used: LatexConcepts Applied:-Video background subtraction using SVD: take the background; reshape to big long vector; use SVD to get the low rank component i.e. the background, and end up with the residual.-Low Rank Matrix/Image completion: put in missing entries into low rank image; solve alternating algorithm using SVD to reconstruct the entire matrix.-Polynomial Fitting using ordinary least squares.-De-noising, low rank matrix + noise utilize SVD at optimal rank and view singular value spectrum Decision theory, setting boundary (probability of correct classification and false classification); ROC curve.-SVD Nearest Subspace classification (linear auto encoder); classifying hand written digits.-Signal and Image un-mixing using ICA & PCA.-Classification using Neural network input/output mapping, stochastic accelerated gradient descent -Procrustes Analysis, for lining up images (translate scale and shift).-MDS, distance to coordinate.-Clustering.-Image In painting and matrix completion utilizing regularized LS. -Finding Synced signals; using SVD.

Canada

Launched CD4.2 (Lincoln Nautilus and Ford Edge) in Oakville Assembly Plant (CD539NXA & U540N/C) VP/TT/PP as Project Manager for PMT 5 (electrical) and PMT 4 (power train).Lead on all open design AIMs as well as DEMs/ VOCF through lateVP, TT, Pre-PP, PP. Worked on 228 AIMs issues supporting them to pending/closure status, appropriately re-binning to other PMTs, SQ, VO. Pended 92 issues, 34 to TT, 51 to pre-PP & PP, and 18 to MP1/2 PJ1. Validated 40 pending AIMs in TT, held weekly meetings with systems. Validated R202 issues & 10 pending calls at VOCF. Completed 3 Functional Quality Checklists (B153, B177, and B172). Led OTG through TT; Resolved 8 obsolete parts & 48 incomplete drawing supplier calls. Got 12 alerts approved and 3 1-pagers signed. Championed issue 3662349. Presented 8 issues to Chiefs/Managers in MCE/WNFBJ1 forum; brought to closure. Traveled to Dearborn, personally reviewed 18 DEMs calls in VOCF/DSC. Worked with D&Rs to get in 8 illustration updates for PP. Involved in TT part sign off at MSC. Signed off on 26 U540 parts and 62 CD539 parts.

United States

Received July 2017 Peer Recognition Award for Exceptional Demonstration of Teamwork under Autonomous Controls & Systems Team.Became liaison between EPE and AV teams. Accurately defined FMEM modes that influence AV torque estimation/power-up power-down/and driver controls. Categorized all ECM/HPCM DTCs from Part 2 spec based on MRC event with AV specific drivability info. Performed on-vehicle and HIL testing for LOS modes in AV architecture documented behavior and identified inaccurate torque capability estimation in LOS-electric modes ….provided temporary solution internally and worked with EPE to put in Etracker to develop more accurate AV specific torque estimation signal. Led meetings and completed comprehensive analysis of driver control usage in HPCM/ECM to better understand how to handle driver controls in AV & supported effort to communicate approach to appropriate SSFTS

United States

Lead CAE for V71X pre-program Range Simulations. Incorporated C727 SCVSP simulation results in current sensitivity tool; Analyzed effects of aero/weight/climate on range & developed an excel doc for the team to perform range calculation on different climate power specs. Created multiple matlab scripts for dump file analysis. Incorporated European drive cycles in sim runs to accurately provide range results to FOE. Performed multiple coast-down, 5-cycle tests in APTL on Ford & competitor vehicles.

United States

Implement/Define/Deliver/Validate. BECM feature management, involved in SW feature release process. Mentored FOI for program delivery/self-improvement and documented full release process for team. Worked on LCP feature model to make a stateflow/simulink change to the fan logic for china model.

Dearborn Michigan

Received Award of Recognition on work to resolve CD Flat Tow Customer Concern. Analyzed customer complaint regarding Gen2 Fusion battery drainage. Performed Dyno testing, collected IDR, PTDiag, Canalyzer data to root cause. Communicated fix to dealer/customer, supported update to owners manual for flat tow procedure.

Greater Detroit Area

Master'S Degree, Autonomous Automotive Engineering, 4.0 Gpa

Master'S Degree, Electrical And Computer Engineering, 3.97 Gpa

Bachelors Degree, Computer Engineering