Himanshu G.

Experience

• 

  SmallDay IT Services Pvt. Ltd.

  May 2016 - Jun 2016

  Summer Intern

  - Worked on development of a transliteration algorithms which converts text written in one language to any other given language.- Optimized the efficiency of the algorithm by Implementing a tries based decision tree which enable it for fast retrieval of result and reduce latency of the requests.

• 

  Citi

  Aug 2017 - Nov 2019

  Trade and Treasury Solutions (TTS) | Recommendation Engine for Data Entry - Developed Recommendation system to assist employees doing data entry from trade documents. - Received Citi Applause Award & Citi Silver Award recognizing individual contributions for various initiative. - Awarded Engineering Excellence Level-2 certification for Following and inculcating Good coding practices Commodities BU | Software Development | Sonar CRM tool - Worked on revamping of Sonar project from Monolithic to Microservice architecture.- Worked in Java Spring Framework, Angular and Oracle SQL.

  • Data Scientist

    Jul 2018 - Nov 2019

  • Analyst Programmer

    Aug 2017 - Jun 2018
• 

  TwoPaksh

  Jan 2020 - Sept 2020

  Tech-Head | Co-Founder

  Technology Head | Co-Founder | Worked on multiple ideas pivoting on Smart-Surveillance-System ◦ Created MTCNN+Face-net based Face Recognition model for indoor communities like Gyms, Office etc ◦ System provided state of art surveillance system, also deriving crucial analytics over generated metrics. ◦ Implemented Transfer-Learning approach on Face-net model to adapt & improve performance over time. ◦ The system ported directly into any IoT camera (browser based), & thus did not cost any extra hardware. Show less

• 

  FogHorn Systems, Inc

  Oct 2020 - Dec 2021

  Data Scientist - II

  Foghorn is recognized as a top AI platform for industrial IoT and edge computing solutions. Worked on multiple computer vision based product which are deployable on edge IoT devices• Blow out Preventer ◦ Developed a Robust Image Augmentation Pipeline, later integrated across other computer vision projects. ◦ Implemented and experimented on Faster R-CNN (with Inception & ResNet-50/101 backbone architectures), performing hyper-parameter optimization during Transfer Learning. ◦ Project received IoT Edge Computing Excellence Award 2021 for implementing edgeAI in remote drilling site• Personal Protective equipment | PPE ◦ Implemented SoTA Copy-Paste Augmentation Technique to improve accuracy on rare classes. ◦ Created Rule Based Accuracy Matrices reporting FrameworkAcquired by Johnson Control Show less

• 

  Johnson Controls

  Jan 2022 - Sept 2024

  DataScientist

  Worked on Development of Industrial IoT based ML applications• Flare Advanced Volume Estimations ◦ Developed mask-segmentation based real time flare monitoring system ◦ System provided real time values on amount of Gas burnt via visual monitoring on multiple stack at once ◦ Co-Author, IEEE Paper Titled "Deep Learning based Flare Image Analytics for Emissions Monitoring at the Edge"• Optimal Start Algorithms for BMS ◦ Development of continuously-Trained & site condition adaptable ML system to predict HVAC start time for smart BMS. ◦ Developed Energy-Saving-Estimation framework using various available sensors data from BMS. ◦ System was expected to have Saved $1Million in a year, which is ∼7% of total incurred bill.• Other Projects: • Developed efficient GPU Utilization Strategy which optimizes concurrent usage of GPU by multiple models deployed in Edgeml© docker container. • CycleGAN implementation for converting synthetic image to real looking image in unsupervised settings. • Selective Target Updation: Training of DL model only for selective output classes w/o affecting other classes. • PPE and Safety Violation: Monitoring safety norms violation based on 20+ categories of objects from the site. Show less

• 

  University of Chicago

  Sept 2024 - now

  Graduate Student