Hey! I am

Harsha

I'm a

About

About Me

I am a Graduate Student at USC pursuing Master's in Electrical Engineering focussing in Data Science, Machine Learning, and Deep Learning with more than 5 months of research experience. I have 3+ years of professional experience in software development and distributed infrastructure. Previous software experience has helped me develop niche skills in web technologies, RDBMS, deployment architectures, common design patterns, and systems design components. I also served different roles in the SDLC, interacted with cross-functional teams, led POCs to develop MVP models for business-critical applications.

Currently, I am working as a student researcher at ISI where I leverage Deep Learning to generate fake images of real people and study similarities with original faces. I am excited to put the skills honed through this software journey to solve real-world use cases. Thereby, I am actively seeking full-time opportunities in Software Engineering, Machine Learning, and Deep Learning.

  • Name: Harsha Raidurgam Venkat
  • Date of birth: December 04, 1993
  • Address: Fremont CA 94536
  • Email: raidurga@usc.edu
  • Phone: +1(213)-284-6653

Download CV

Education

2020-2021

Master of Science (MS) in Electrical Engineering

University of Southern California (USC)

Focussed on developing competency in Data Science and Machine Learning.

Coursework: Analysis of Algorithms, Programming Systems Design, Pattern Recognition, Parallel and Distributed Computation, Deep Learning, Machine Learning.

2011-2015

Bachelor of Engineering (BE) Telecommunications

Visvesvaraya Technological University (VTU)

Coursework: Data Structures and Programming with C++, Operating Systems, Embedded Systems.

Experience

Jun 2021 – Present

Graduate Student Researcher

Information Sciences Institute (ISI), Marina del Ray, CA

  • Conduct AI research on GAN inversion, extract facial embeddings using SOTA models, optimize projected face to be as similar to target face, and evaluate facial similarity metrics. Critically evaluate model performance by weights pruning and quantization of Resnet, Mobilenet CNN models with PyTorch.
  • Develop a configurable distributed deep learning framework to optimize scheduling and projection time for ~1000 jobs scheduled with SLURM. Built with PyTorch Lightening, Ansible, Kubeflow.
  • SOTA models: VGG-Face, Google FaceNet, OpenFace, Facebook DeepFace, DeepID, ArcFace, and Dlib and PyTorch models from Face-X zoo.
  • Submitting a technical paper for CVPR, 2022

Jan 2019 – Oct 2019

Senior Software Engineer

Torry Harris Integration Solutions (THIS), Bagalore, India

  • Automated functional testing for 150+ REST APIs/SOAP services and reduced manual testing time by 75%.
  • Engineered a smart log analyzer app, drastically minimized raw log analysis and issue debugging time by 68%.
  • Conducted a POC with a team of 4 to design an E-Commerce portal with Commerce Tools, Springboot microservices backend, and GraphQL interface. Onboarded the project, and added $100k revenue in the first cycle.
  • Part of migration team to strategize, plan and migrate APIs/web services from SOA to Microservices architecture.

Apr 2017 – Dec 2018

Software Engineer

Torry Harris Integration Solutions (THIS), Bagalore, India

  • Developed business-critical e-commerce solutions for purchase & payments, subscriptions, product provisioning, and finance-related workflows that received over 100M+ hits every month. Employed Apache Kafka, RabbitMQ, WSO2 EI, and Springboot microservices to develop end-to-end workflows
  • Developed hive scripts to extract 18M+ records from 14+ data sources every day to compute API statistics. Scripted a high visibility custom Analytics Dashboard with Elasticsearch, Kibana to aid stakeholders, business partners to uncover KPIs for decision studies and brought down annual costs by 200,000€ (by 32%)
  • Developed API Explorer portal to enable consumer onboarding. Integrated front-end with “Try it” functionality using Swagger-UI and generated sample client codes to consume APIs in Java, Javascript, PHP, CURL, and Python
  • Fixed critical platform vulnerability issues on 30+ AWS VPC applications and mitigated DoS, SQL Injection attacks
  • Paired with a teammate to migrate 90+ application servers from On-Premise Data Center to AWS regions across EU, US, and China. Developed Ansible scripts, devised migration strategies, and effectively migrated infrastructure with minimum downtime.
  • Sped up data center to cloud migration by 31% using a constrained, and parallelized data transfer algorithm and cut down data transfer time by 64%
  • Redesigned pagination for legacy application to display ~5M records 3x times faster by reindexing and using SQL offsets
  • Drove three cross-functional teams to address long-standing issues to closure and decreased backlog by 63%

Mar 2016-Mar 2017

Associate Software Engineer

Torry Harris Integration Solutions (THIS), Bagalore, India

  • Built a SpringMVC app for 10,000+ B2B partners and B2C users to create, download, and install proprietary licensing software. Partnered with AWS Cloud Ops, Firewall teams to resolve TCP packet loss issues and raised API reliability by 28%.
  • Redesigned pagination to display ~5M records using data tables, PostgreSQL offsets, caching mechanism, and achieved 3x faster page load time.
  • Award: Best Performer for enhancing health check monitoring efficiency of 90+ application servers through automation and reducing required man-hours by 98%; promoted 6 months ahead of schedule

Awards

2016-2017

Best Overall Performer

Torry Harris Integration Solutions

Best Performer for enhancing health check monitoring efficiency of 90+ application servers through automation and reducing required man-hours by 98%; promoted 6 months ahead of schedule.

2015-2016

'Rookie Rockstar' - Best Fresher

Torry Harris Integration Solutions

Designed, developed and maintained Spring MVC app for 10,000+ B2B partners and B2C users to create, download, and install proprietary licensing software.

2015-2016

Top 200 among 6000+ teams

INGENIUM by Quest Global, India

'Advanced Wheelchair Prototype’ was shortlisted among 6000+ teams throughout India and was among the top 200 teams shortlisted for the final round.

2015-2016

Top 3 among 30+ teams

IESA-TiE IoT in Smart Healthcare Awards for Academia

Top 3 among 30+ teams across India to present the project ‘Advanced Wheelchair Prototype’ in a contest held by India Electronics and Semiconductor Association (IESA) and The Indus Entrepreneurs (TiE).

2015

Best Innovative Project

Sir MVIT, Bangalore, India

Awarded best innovative final year thesis project for ‘Advanced Wheelchair Prototype’.

Projects

Projects



Machine Learning

Selected Projects from Machine Learning

Vertibral Column data Classification

KNN | CLASSIFICATION | GRID SEARCH

Movie Recommendation System

COLLABORATIVE FILTERING | CONTENT-BASED FILTERING

Facebook comment volume

ENSEMBLE | XGBoost | ADABOOST

Acoustic Data(MFCCs) Classification

CLUSTERING | MULTI-CLASS CLASSIFICATION | SVM | SMOTE

Linear Regression Analysis

OLS | LASSO | RIDGE

Time Series Analysis

ARIMA | SHRNIKAGE | FEATURE EXTRACTION | LOGISTIC REGRESSION | CLASSIFICATION


Deep Learning

Selected Projects from Deep Learning

Handwritten Digits Classification

MLP | CLASSIFICATION | CNN | DATA VISUALIZATION

Image Synthesis using Conditional GANs

CNN | IMAGE SEGMENTATION | GAN | IMAGE SYNTHESIS

Polyvore Fashion data Classification

CNN | CLASSIFICATION | SIAMESE

Fashion MNIST Classification

CNN | CLASSIFICATION | SGD

Human vs Computer Random Sequences

PCA | MLP | CLASSIFICATION


Web Development

API Documentation with Redoc

ReactJS | SWAGGER | Redoc | HTML | CSS

WebRTC Video Conferencing App

WebRTC | GAN | FLASK | SocketIO | OpenVINO

Smart Log Analyzer

ReactJS | MATERIAL UI | ExpressJS | WEBPACK

I'm Available for hire

Would like to start a conversation?

Contact me

Contact

Contact Me

LinkedIn

View my LinkedIn profile

Contact Number

+1(213)-284-6653

Email Address

harsha.rv67@gmail.com

Website

View my GitHub repos