Phani Kumar

System Design and Core Backend Implementation

Django Framework: Developed a robust backend using Django and Django REST framework, integrating an ASGI app with full async capabilities through Channels.

Database Design: Utilized PostgreSQL with JSONB fields for efficient storage and retrieval of user chats, and employed Redis for in-memory caching to enhance performance.

Asynchronous Processing: Implemented Celery for managing background tasks and periodic synchronization of Redis chat data to PostgreSQL, with Redis serving as the message broker between Channels and Celery.

Hosting and Deployment: Deployed the application on AWS EC2, utilizing Nginx as a reverse proxy server and Daphne as the ASGI server to handle asynchronous requests.

Object Storage: Integrated AWS S3 for scalable object storage solutions.

Notification System: Incorporated Google Firebase to handle real-time notifications and updates.

Dashboard: Developed interactive dashboards and reports using Streamlit Framework

Developed Complex Strategies: Designed and implemented advanced backtesting strategies using Pandas and Numpy, aligned with stakeholder requirements.

Applied Object-Oriented Programming: Utilized OOP principles to construct and integrate complex financial models and strategies for various instruments.

Executed Historical Data Analysis: Conducted backtesting on seven years of historical data for NSE stocks and indices, emphasizing performance and accuracy.

Optimized Performance: Leveraged multiprocessing and multithreading to enhance the speed and efficiency of backtesting processes.

Focused on Design and Logic: Excelled in both the architectural design and the development of intricate logic, ensuring effective strategy evaluation and implementation.

SparkRadius: Software Developer

Project Overview: Developed an OSS/BSS (Operations Support Systems/Business Support Systems) application utilizing Python Django for robust web development and MySQL for relational database management. Implemented a microservice architecture to ensure scalability and maintainability across various system components.

Database and Caching:

• MySQL: Used for structured data storage and management, ensuring reliable and efficient handling of operational and business data.
• Redis: Employed as an in-memory data store to enhance performance through caching and to support fast data retrieval and processing.

Asynchronous Processing:

• Celery: Integrated for task management and background job processing, allowing asynchronous execution of tasks and improving overall system responsiveness.
• Flower: Utilized for monitoring and managing Celery workers, providing real-time insights into task execution and worker performance.

Event Streaming:

• Kafka: Implemented for real-time data streaming and messaging, facilitating reliable and scalable communication between microservices and ensuring the efficient handling of high-throughput data.

Microservice Architecture:

• Designed and developed the application with a microservice approach, enabling independent deployment and scaling of services, enhancing system flexibility, and improving fault tolerance.

Ecommerce Scraper Project: Lead Developer

FastAPI Framework: Developed high-performance scraping endpoints using FastAPI, designed to efficiently handle multiple Indian eCommerce sites.

Concurrent Processing: Implemented multiprocessing to enable simultaneous processing of input URLs, addressing challenges associated with sharing complex objects like Pandas DataFrames and Selenium WebDriver, which are not natively serializable.

Advanced Techniques: Utilized multiprocessing queues, values, and events to ensure robust and concurrent data scraping, adhering strictly to D.R.Y (Don't Repeat Yourself) principles

Web Scraping Tools: Integrated Selenium for dynamic web interactions and BeautifulSoup (Bs4) for parsing HTML content, enhancing the scraping process.

Qspeech – Automatic Speech Recognition (ASR):

• Developed an ASR system for converting speech to text using DNNs and RNNs.
• Integrated emotion detection with the valence-arousal technique to improve ASR in emotionally sensitive scenarios.
• Created algorithms to identify silence and enhance audio data quality for ASR training using DL techniques.
• Expertise in AI, ML, DL, and audio signal processing for feature extraction and noise handling.
• Delivered a high-performance ASR system with accurate transcription and emotion-aware capabilities.