Making Machines Learn
Pushing Theory to Practice
For the past year and a half, I have been working as an ML engineer trying to automate and deliver products using Machine Learning for the many tedious processes that we as humans go through every day. My primary domain of work is as mentioned.
I work in various Computer Vision use cases like Facial Recognition, Object Detection, Object Tracking, Eye gaze estimation, Face Angle estimation, Liveness Detection, etc.
Over the years, I have developed a keen interest in making our cities smarter using Artificial Intelligence.
I push deep learning models to production using NVIDIA’s TensorRT & DeepStream alongside TensorFlow and PyTorch.
I love giving back to the ML community and sharing knowledge with others. I try to help out as many people as I can.
Some Interisting Facts
I love trekking and the feeling that comes after climbing a mountain. On a side note, I am a Manchester United fan, and I love reading books.
Some Words About Me
My Awesome Story
I started with Linear Regression, and now I deal with training deep neural networks. I try to create a synapse between vision and the machine to create a tiny virtual brain. To simplify, I work in Computer Vision.
Senior ML Engineer – Guise AIMay 2020 - Present
- Leading and guiding our Machine Learning team in the effective use of AI and data in Computer Vision use cases.
- Formulating machine and deep learning approaches while paying attention to business metrics, designing features from the rich data available from many sources, training, evaluating, and deploying models.
- Conducting research and case studies on leading-edge technologies to make determinations on the probability of implementation.
- Working cross-functionally to define problem statements, collect data, build analytical models, and make recommendations.
ML Engineer – Guise AISeptember 2019 - May 2020
- One of the lead developers in the Machine Learning team with a primary focus on delivering solutions to customers both as a service and a product.
- Formulated Machine Learning and Deep Learning approaches for various use cases in a smart city, retail shop, petrol pump, golf court, and more.
- Worked on improving the current face recognition system and achieved a state-of-the-art accuracy of 99% (Embedding loss) on Labelled Faces in the Wild dataset.
- Developed end-to-end pipelines for Computer Vision use cases that leverage NVIDIA’s TensorRT and DeepStream for fast inference on dedicated hardware, cloud services, and edge devices.
- Implemented eye gaze estimation and face angle estimation to gather insights from a person’s face.
- We engineered an end-to-end pipeline to extract clothing and apparel from videos and images, leveraging our in-house segmentation algorithm.
- Generated feature-rich datasets for face detection, face recognition, liveness detection, facial attributes classification, object detection, object tracking, pose estimation, license plate detection, and OCR; and training, evaluating, and deploying models for the same.
Software Engineer – miniOrangeJuly 2019 - September 2019
My Developer's Skills
I work with multiple frameworks for deep learning and machine learning, using Python and C++. I am also interested in assembling machines for deep learning and gaming.