Developing LLM based solutions to increase Revenue

Develop and Deploy Computer Vision based warehouse safety solutions for Edge Compute

Lead Engineer

• Led multiple projects from POC to production resulting in cost savings of over $10M PA

• Worked on a wide array of Multi-Modal deep learning challenges including Classification, Segmentation, Object Detection, Pose Estimation using CNNs, Transformers, VLMS.

• Successfully tackled non-traditional modalities like raw Laser Waveform, Sonar and Point Clouds.

• Developed in-house machine learning training, inference and metrics libraries catered towards special data modalities.

• Built tools for business users to interacts with machine learning systems to perform exploratory data analysis, label data and train models.

• Submitted 3 patents for pioneering machine learning solutions, demonstrating innovation and expertise in the field

• Designed and implemented cutting edge highly scalable inference solutions on AWS cloud infrastructure using Amazon SageMaker,

EC2, AWS Lambda, StepFunctions, ensuring efficient model deployment and enhancing customer experience

• Collaborated closely with cross-functional teams across the Americas and Europe regions, working with product managers and business

users to identify and address machine learning challenges

• Customized and deployed a tailored version of the CVAT data labeling platform to meet in-house user requirements

• Led intiatives for new innovations, prototypes, user trainings and presented demos and trainings for subject matter experts and leaders

• Mentored junior members of the team, providing guidance on technical and career development

Founding Engineer

• Developed a system to measure Heart Rate Variability (HRV), a critical biomarker for tracking physiological stress and athletic recovery,

by leveraging the accessibility of standard smartphone cameras and face segmentation

• Established a baseline for the project by first engineering an HRV algorithm that utilized a contact-based, finger-placement method with

the phone’s camera and flash.

• Engineered a remote photoplethysmography (rPPG) pipeline that identified facial landmarks, isolated forehead and cheek regions, extracted a pulse signal from RGB pixel data, and calculated HRV from R-R intervals, achieving results within a 7% tolerance of dedicated

chest and wrist-worn sensors.

Founding Engineer

• Initiated and led machine learning initiatives as the company’s first ML engineer

• Developed cutting-edge deep learning models for road feature classification and segmentation in aerial and satellite imagery, utilizing PyTorch and QGIS. Improved the baseline by 30%

• Managed end-to-end ML pipelines, covering data acquisition, preprocessing, model containerization, and deployment on Google Kubernetes Engine

• Oversaw recruitment efforts, hiring both full-time staff and interns specialized in software and ML workflows