Hey! 👋🏾 I'm Surya Krishnan

At Johnson & Johnson , I built a deep learning and computer vision based defect detection system that identifies and highlights defects and impurities present in products produced in J&J manufacturing plants to prevent defective products from reaching consumers.

As part of the defect detection system, I built unsupervised visual anomaly detection, unsupervised image retrieval and object detection modules. Several digital image processing techniques were also implemented in the data processing pipeline.

At Johnson & Johnson, I helped implement a computer vision based surgical tray inspection tool that helps medical equipment manufacturers ensure that the right set of components are present in the appropriate surgical trays. Automating this inspection through the tool saved a lot of time and enabled the users to focus on other critical processes in the business.

I also built a sales forecasting engine that helped automate and streamline Johnson and Johnson’s demand planning pipeline for their ecommerce division in China.

At Safeo , I built a computer vision based platform that helps companies get insights about employee and customer safety at their workplaces.

Mask detection (face detection and object detection), Social Distancing Monitor (object detection and depth estimation from a single image), and a state of the art face recognition based attendance system are some of the key modules I designed and built.

I also designed and implemented the backend architecture (database schema, API design and a queue system) to handle the huge amounts of video data that needs to be processed while maintaining real-time performance.

At SIO , I worked with Marine Biologists to build a vision system that continuously monitors the ocean for poisonous algae presence and alerts the concerned authorities when there are lots of them.

I worked with the biologists to curate the dataset of poisonous and benign algae images. I also built the neural net classifier that takes in underwater images of algae and predicts whether they are poisonous.

Betoken is a meritocratic and decentralized crypto assets hedge fund built on the Ethereum blockchain.

At Betoken, I built the front end for the dApp using AngularJS and used the web3 library to interact with Betoken’s smart contracts on the Ethereum blockchain.

I also handled strategy, fundraising and community outreach for Betoken.

Locbit is an IOT company that aims to optimize energy consumption in buildings. At Locbit, I built machine learning models that predict the client buildings’ energy usage for any given day.

At Locbit, I also worked on another initiative whose goal was to drive all of the energy-usage optimizations in the client buildings through a reinforcement learning algorithm.