Progyan Gupta

Software Developer – Computer Vision & AI

Software Developer focused on AI and real-time video analytics, with additional experience in NLP and deploying machine learning models. Experienced in delivering scalable, low-latency streaming solutions with high system reliability, supported by streamlined CI/CD processes and performance-driven system optimization.

progyan@localhost:~$ cat skills.txt

PythonGStreamerDeepStreamKafkaTritonPytorchRGoLangDockerAzureCICD

Scroll Down

Work Experience

Software Developer – Computer Vision

Nov 2022 – Present · Melbourne, AU

Developing real-time video analytics pipelines using DeepStream, Kafka, and Triton for scalable, low-latency AI applications.

Key Contributions:

Built and optimized AI pipelines for object detection, tracking, and re-identification across distributed systems
Integrated DeepStream and Triton for efficient multi-model inferencing
Designed Kafka-based event streaming and Kafka Connect pipelines for real-time processing
Maintained 90%+ test coverage with robust CI/CD pipelines and automated testing
Handled RTSP video streams and ensured continuous video handling in production

DeepStreamNVIDIA TritonCudaKafkaKafka ConnectCI/CDRTSPPythonDocker

Data Science Intern – NLP

Jan 2021 – Feb 2021 · Remote

Applied transfer learning to fine-tune state-of-the-art NLP architectures for sentiment analysis, and benchmarked their performance

Key Contributions:

Deployed a transformer-based model using Hugging Face for customer feedback classification
Benchmarked multiple pre-trained models such as BERT, RoBERTa, and XLNet for optimal performance
Streamlined deployment workflows for scalability using open-source NLP APIs
Gained practical experience in production-ready model evaluation and end-to-end machine learning deployment

Hugging FaceTransformersBERTRoBERTaXLNetPythonScikit-learnPandas

Impact & Achievements

System Performance

Real-Time Inference

• Integrated multi-model object detection pipelines to improve latency and accuracy
• Achieved low-latency processing of RTSP streams at production scale
• Boosted inferencing throughput using NVIDIA Triton Inference Server

Model Optimization

• Evaluated and optimized AI models like PeopleNet and YOLO for real-world video applications
• Deployed object tracking and re-identification at scale
• Reduced system overhead while maintaining high detection accuracy

Infrastructure & DevOps

Streaming Infrastructure

• Designed Kafka-based event pipelines for real-time event streaming
• Integrated Kafka Connect to stream analytics to external endpoints
• Managed distributed RTSP ingestion with seamless video handling

CI/CD & Reliability

• Achieved 90%+ test coverage through automated testing
• Built robust CI/CD pipelines for production deployments
• Enabled reliable updates with zero-downtime rollouts

Cross-Functional Development

NLP & ML Deployment

• Deployed sentiment analysis models using Hugging Face API
• Benchmarked transformer models for real-world feedback categorization
• Improved classification accuracy and deployment scalability

Cross-Stack Contribution

• Delivered solutions across backend, infrastructure, and streaming systems
• Contributed to both development and architectural decisions
• Solved complex technical challenges in high-performance environments

progyan@localhost:~$ cat connect.txt

Let's Connect

We all build, learn, and grow — sometimes better together. If our paths align, let’s talk.

progyan@localhost:~$ location --current

Melbourne, AUS

progyan@localhost:~$ contact --email

contact@progyan.dev

progyan@localhost:~$ cat resume.pdf

Download Resume

progyan@localhost:~$ ls ./social-links

GitHub

@Progress04

Progyan Gupta