Profile

Sai Kiran

About Me

Python & Deep Learning Enthusiast with a continuous learning passion for Engineering, focused on creating practical and impactful real-world applications. Experienced in computer vision, efficient image data pipelines, model training, and automation of Python-based SDK workflows, developing REST APIs. Continuously exploring advancements in AI experimenting with prompting, model evaluations, and building mini POCs for research, quick solutions, and personal use-cases. Proficient in Python, deep learning frameworks, databases, and web technologies, complemented by prompt engineering for intelligent interactions. Curious and adaptable, ready to explore diverse roles, often transforming innovative ideas into functional prototypes using GenAI tools like Cursor IDE and ChatGPT, deploying them on accessible platforms for real-world utility.

Projects

Retail Autonomous Store

A Cashier less store based on walk-in and walk-out strategy where customers simply walk-in and pick-up the necessary items and walk-out where the bill gets generated in a automated-way.

Tech-Stack: Python, Dataloop, computer vision, Deep learning, Opencv, numpy, augmentations, Pillow, MinIO storage system, Linux OS, Flask, CNNs

Retail Kirana Store Portal - Personal project

Developing a web application for managing retail store operations with modules for stock inventory tracking. Current features include stock management. Upcoming modules under development include sales data recording, and date-range-based reporting, enhanced sales analytics, credit management for tracking customer credit, and a user authentication system for secure access.

Tech-Stack: Flask, SQLiteDB, HTML, CSS, JavaScript, and Git for version control.

Road Transport Accident Severity Prediction

Prediciton of injuries and the severity of injury for people who met in road accidents

Tech-Stack: Python, pandas, numpy, seaborn, matplotlib,scikit-learn, Streamlit

Professional Profile

Work Experience

Python Software Developer

Limasoft Pvt. Ltd

January 2024 - Present

Projects:
    An AI-powered conversational agent, similar to ChatGPT, designed for supply chain management. It allows users to input natural language queries for summaries or contextual insights. The system employs a Retrieval-Augmented Generation (RAG) approach using ChromaDB as the vector store to fetch relevant supply chain data, which is then processed by the LLM to generate accurate, actionable responses in real time.
Roles:
Python Developer:
  • Implemented ChromaDB for efficient and accurate data retrieval through query operations.
  • Utilized GPT models for contextual sentence generation, supporting tasks like code generation and information retrieval.
  • Optimized performance through batch API processing, enhancing efficiency in handling multiple tasks
  • Designed and developed REST APIs for diverse use cases, ensuring robust and scalable solutions.
  • Conducted in-depth research on distributed database systems like Dremio and Trino and developed integrated code to connect with Trino for executing federated queries (cross-db queries), establishing it as a viable alternative to PySpark.
  • Developed a Python-based IDE leveraging the MONACO editor for app integration, enabling script creation with predefined conditions and execution using the runpy library
  • Managed CI/CD pipelines using Jenkins, ensuring continuous integration via gitlab

Tech Stack: Python, Git, REST API Development, Flask, DBMS(PostgreSQL, MariaDB, MSSQL, oracle, trinodb, clickouse DB), GitLab, Jenkins, Natural Language Processing, gpt-3.5, gpt-4o-mini,

Systems Engineer - Autonomous Retail Store

Infosys

August 2021 - December 2023

Projects:
    Autonomous Retail Store: A cashier-less store system using a walk-in and walk-out strategy, enabling automated billing upon customer exit.
Roles:
Python Developer, Computer Vision Engineer
  • Developed a Python-based SDK pipeline for data management, labeling, and model training, producing ONNX format outputs.
  • Created a Flask-based UI for image augmentation using the Albumentations library.
  • Designed and developed a synthetic data generation pipeline with a data scientist, reducing data capture and annotation efforts.
  • Built an auto-annotation pipeline, improving annotation efficiency by 30%.
  • Evaluated various models for use-case suitability, focusing on body-pose estimation and triplet loss classification.
  • Trained and evaluated object detection models for real-world applications.
  • Achieved superior object detection performance by combining synthetic and manually collected data
  • Specialized in body-pose estimation and triplet loss classification models
Data Annotator & Engineer:
  • Managed and processed thousands of images for computer vision model training and evaluation.
  • Managed and been part of Annotations team, by annotating images in various categories as per the requirement for the object detection model training and evaluations for real-time model performance

Tech Stack: Python, Computer Vision, Deep Learning, OpenCV, PyTorch, Flask, ONNX

Skills

Programming Languages & Libraries

  • Python
  • PyTorch
  • OpenCV
  • Numpy
  • Pandas
  • Matplotlib
  • PIL
  • Flask
  • REST API Development

Data Processing & Analysis

  • Image Processing
  • Computer Vision
  • Data Structures and Algorithms
  • SQL (Beginner)

Machine Learning & AI

  • Convolutional Neural Networks (CNNs)
  • Transformers
  • Prompting with LLMs (GPT-3.5, GPT-4)
  • OpenAI API
  • NLP models (BART, BERT)
  • Stable Diffusion (Basic)
  • Azure Custom Vision
  • DeepStream (Basic)
  • ONNX

DevOps & Tools

  • Linux
  • Git
  • Jenkins
  • IoT (Basic)
  • Dataloop

Certifications

IBM Applied AI Professional Certificate

IBM Applied AI Professional Certificate

Coursera

Feb 2023

THDVZ36JTYA7

Mega Guided Projects Certificate

Mega Guided Projects Certificate

The Machine Learning Company

Feb 2023

MGP300200

Get in Touch

Connect with me via email or LinkedIn, or drop a message below!

Work Experience

Python Software Developer

Limasoft Pvt. Ltd

January 2024 - Present

Projects:
    An AI-powered conversational agent, similar to ChatGPT, designed for supply chain management. It allows users to input natural language queries for summaries or contextual insights. The system employs a Retrieval-Augmented Generation (RAG) approach using ChromaDB as the vector store to fetch relevant supply chain data, which is then processed by the LLM to generate accurate, actionable responses in real time.
Roles:
Python Developer:
  • Implemented ChromaDB for efficient and accurate data retrieval through query operations.
  • Utilized GPT models for contextual sentence generation, supporting tasks like code generation and information retrieval.
  • Optimized performance through batch API processing, enhancing efficiency in handling multiple tasks
  • Designed and developed REST APIs for diverse use cases, ensuring robust and scalable solutions.
  • Conducted in-depth research on distributed database systems like Dremio and Trino and developed integrated code to connect with Trino for executing federated queries (cross-db queries), establishing it as a viable alternative to PySpark.
  • Developed a Python-based IDE leveraging the MONACO editor for app integration, enabling script creation with predefined conditions and execution using the runpy library
  • Managed CI/CD pipelines using Jenkins, ensuring continuous integration via gitlab

Tech Stack: Python, Git, REST API Development, Flask, DBMS(PostgreSQL, MariaDB, MSSQL, oracle, trinodb, clickouse DB), GitLab, Jenkins, Natural Language Processing, gpt-3.5, gpt-4o-mini,

Systems Engineer - Autonomous Retail Store

Infosys

August 2021 - December 2023

Projects:
    Autonomous Retail Store: A cashier-less store system using a walk-in and walk-out strategy, enabling automated billing upon customer exit.
Roles:
Python Developer, Computer Vision Engineer
  • Developed a Python-based SDK pipeline for data management, labeling, and model training, producing ONNX format outputs.
  • Created a Flask-based UI for image augmentation using the Albumentations library.
  • Designed and developed a synthetic data generation pipeline with a data scientist, reducing data capture and annotation efforts.
  • Built an auto-annotation pipeline, improving annotation efficiency by 30%.
  • Evaluated various models for use-case suitability, focusing on body-pose estimation and triplet loss classification.
  • Trained and evaluated object detection models for real-world applications.
  • Achieved superior object detection performance by combining synthetic and manually collected data
  • Specialized in body-pose estimation and triplet loss classification models
Data Annotator & Engineer:
  • Managed and processed thousands of images for computer vision model training and evaluation.
  • Managed and been part of Annotations team, by annotating images in various categories as per the requirement for the object detection model training and evaluations for real-time model performance

Tech Stack: Python, Computer Vision, OpenCV, Pillow, PyTorch, Linux, Albumentations, DataLoop, Deep Learning, Rembg, Flask, ONNX