Harshit Sharma

Experience

• 

  Elixir Technologies

  Jun 2019 - Aug 2019

  Project Intern

  Worked along with Web Technologies for multiple Projects, using Adobe XD for wireframing and development using ASP.NET.

• 

  KoiReader Technologies

  Apr 2020 - Jul 2020

  Computer Vision Intern

  • Created POC for an Automated Reports generation and mailing service.• Researched and implemented multiple NLP tasks using various Python libraries. Created internal knowledge base for the same.• Created easy access NLP libraries for integration with internal OCR API and other data pipelines.

• 

  Bay Leaf Digital

  Jun 2021 - Dec 2021

  Data Engineer

  • Helped create Novel Data Analytics Pipelines in the company.• Created frameworks for comparison of open-ended metrics from variety of Data Sources.• Helped create complex visualisations as part of the Amplytics Product POC.

• 

  NeurIOT Labs

  Dec 2021 - now

  Senior Data Scientist

  VisionBee (Computer Vision Product) • Leading a team of 6 (Data Scientists and Developers) to create a multi-domain, scalable vision-based analytics application. • Spearheaded the architecture and technical design for VisionBee. • Wrote resource-efficient libraries for generic ETL pipelines from first-principles, capable of scaling reliably, easy to deploy across many devices with plug-n-play features.Automated Pipeline for tracking and entry of vehicles, trucks and trailers for one of the top logistics company. • Defined automation scope with the state-of-the-art in Deep Learning (CNNs). • Achieved 95%+ mAP in detection and identification of relevant entities. • Achieved 98%+ accuracy in Vehicle Number related OCR. • Delivered real-time data events using state of the art messaging systems, with near 99% uptime for the end solution.An end-to-end Retail Customer Analytics Pipeline and Dashboard. • Trained models to detect and track customers in retail (CCTV) environments. • Achieved detection near 99% mAP and ~80% tracking accuracy on low-powered IOT devices running on Edge. • Created a POC dashboard in Python using Dash/Plotly to create end-to-end insights from rawdata to final customer centric views. • Created software architecture with longevity and resilience, capable of handling similar tasks. • Setup live real-time pipeline on edge, handled multiple data driven streams for calculating customer metics in real-time. • Scaled the solution from edge to cloud. Setup large-scale (multi-terabyte) production and consumption media and data pipelines on Google-Cloud Platform. Show less