1. Designed and Implemented Perception Training Framework.

2. Developed automation scripts for automatic generation of code based on configurations to eliminate manual errors and reduce software turn around time by order of weeks.

3. Deployed unit test and integration test pipelines for ensuring quality and compliance of developed code.

4. Assisted cloud testing deployment on aws using prefect and docker on SIL platform for high performance parallel testing of million scenarios.

5. Dockerized applications and build environments for deployment and builds.

6. Established repositories, jenkins pipeline, integrated unit test frameworks to meet production quality.

Led a team of 20+ engineers in the research and development of advanced computer vision systems for autonomous vehicles.

Designed Deep learning training framework on top of pytorch to commonize libraries on loss functions, detectors, backbones etc. Integrated tools like MLflow/W&B for model tracking.

Lead the hardware requirements and design for the domain ADAS embedded controller for L2+ application.

Established realistic expectations on software feature deliveries for the upper management. Increased visibility of the upper management in the development activities by deploying scrum and traceability.

Designed a deep learning-based lane detection system, improved lane detection accuracy by 35%.

Lead sensor fusion algorithm development which used JPDAF, extended kalman filtering, track management algorithm, plausibility checks, existence probabilities etc.

Oversaw the end-to-end deployment of a sensor fusion system, integrating LiDAR, radar, and camera data to enhance environmental perception.

Established repositories, enforced git flow using regression testing, jenkins CI CD pipeline, integrated unit test frameworks, integrated QAC tools for static analysis to meet production quality.

Mentored junior engineers and conducted technical training sessions, improving team skill sets and productivity.

Collaborated with external partners and academic institutions on cutting-edge research projects, resulting in invention disclosures.

Established interview process, hackerrank coding questions and pipeline to filter highly qualified candidates and built a cross functional team.

Researched state of the art neural networks architectures such as transformers, BERT and GPT. Trained and deployed 3d object detection algorithms such as PointNet, VoxelNet, segmentation transformers on a GPGPU.

Developed and optimized computer vision and 3d vision algorithms for real-time semantic segmentation of roads, lanes and object detection algorithms

Led the implementation of architectures like RESNET-50 and UNET for Semantic segmentation problems.

Conducted rigorous testing and validation of computer vision systems, ensuring high reliability and performance in diverse driving conditions.

Showed benefits of recurrent connections in convolutional neural networks (CNNs) at different layers, compared feature maps extracted from CNNs demonstrated how feature maps in CNNs can boost certain features and how segmentation tasks could aid object detection tasks .

Collaborated with cross-functional teams to integrate AI models into the overall autonomous driving stack.

Developed algorithms for Lidar ground plane estimations and clustering points using point nets.