Skip to content
View harsha148's full-sized avatar
๐ŸŽฏ
Focusing
๐ŸŽฏ
Focusing

Highlights

  • Pro

Block or report harsha148

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this userโ€™s behavior. Learn more about reporting abuse.

Report abuse
harsha148/README.md

Hi there, I'm Sai Harsha ๐Ÿ‘‹

๐ŸŽ“ I'm a Master's student in Computer Science at Rutgers University with 3 years of prior industry experience as a Software Engineer.
๐Ÿ’ก Passionate about backend systems, cloud computing, and AI-driven solutions.
๐Ÿš€ I enjoy building scalable software systems, experimenting with intelligent automation, and turning ideas into impactful projects.


๐Ÿ’ผ Experience

  • Backend Engineer Intern @ Eulerity
    Focused on backend services integrating with social media APIs, integrating Meta webhooks for real-time tracking, and configuring scalable log archival strategies on Google Cloud that reduced costs significantly.

  • Research Assistant @ Rutgers University
    Built a computer vision pipeline with YOLOv8 to analyze chemical test strips for coral reef research, reducing analysis time by 90% and improving accuracy in predicting chemical concentrations.

  • Software Engineer @ Wells Fargo (Capital Markets Division)
    Worked on full stack development for payment systems processing FED/SWIFT wire transactions worth $1B+ in daily transfers.

  • Software Engineer Intern @ OYO (Oravel Stays Pvt Ltd)
    Enhanced monitoring of services by building Prometheus-based API health reporting and deployed serverless infrastructure on AWS Lambda for scheduled monitoring.

  • Software Engineer Intern @ Nucleus Software
    Designed a financial chatbot using Rasa with speech and OCR features, improving accessibility and reducing manual data entry time.


๐Ÿ“Œ Projects

  • Intelligent Job Search Assistant
    Built using LangGraph and DeepAgents, this assistant automates job discovery and email communication. One sub agent integrates with LinkedIn MCP for fetching job insights, while another manages Gmail communications, streamlining the application process with an agentic AI.

  • TestStripDX: Object Detection of URS-10 Strips
    Created an very affordable alternative to expensive lab equipment by training YOLOv8 models to detect URS-10 test strip components and predict chemical concentrations using OpenCV-based color analysis, cutting costs while increasing efficiency.

  • Twitter Search
    Designed a high-performance Twitter search engine using FastAPI, PostgreSQL, MongoDB, and Redis. Built data pipelines for 100k+ users and 135k tweets, optimized queries with advanced indexing, and achieved sub-millisecond response times with caching.

  • Web Crawler
    Developed a modular web crawler with a strong emphasis on object-oriented design and clean coding practices. The system was built with extensibility in mind, making it simple to integrate new scraping and parsing features without disrupting the existing framework.

  • Notification System
    Architected a scalable and reliable microservices-based notification system inspired. Implemented Kafka pub-sub architecture with FastAPI services for notification scheduling, user preferences, and delivery. Containerized with Docker and tested with load simulation, it demonstrated high throughput and low-latency under stress.

  • Sentence-Level AI-Generated Text Detection
    Explored the detection of AI-generated vs. human-written sentences using transformer-based models. Designed experiments around sentence embeddings, classification thresholds, and benchmark datasets to improve detection reliability.

  • Brave Bot (AI Decision-Making Bot)
    Developed a grid-based rescue bot using policy iteration, value iteration, and deep neural networks in a MDP based grid. The bot learned optimal strategies to herd a crew member to safety, reducing evacuation time from 236 steps to ~25 steps, demonstrating reinforcement learning in action.

  • ChatFusion: Real-Time Chat Application
    Built a full-stack messaging system with Java, Spring Boot, React, WebSocket, and AWS. Achieved <50ms latency using webhooks for real-time messaging, integrated LangChain + OpenAI as a chatbot assistant, and deployed with AWS EC2, RDS (MySQL), and S3 hosting.

  • AI-NewsHub
    Developed a news aggregation platform using Python, Django, BeautifulSoup, and Streamlit. Aggregates and displays 100+ daily AI-related articles from YouTube, newsletters, blogs, and conference feeds, serving as a one-stop AI insight hub.


๐Ÿ› ๏ธ Skills

Languages: Python, Java, C++, JavaScript, TypeScript, C#
Frameworks & Tools: Spring Boot, Django, FastAPI, Flask, React, Next.js, LangChain, DeepAgents, Rasa
Cloud & DevOps: Google Cloud Platform, AWS, Docker, Kubernetes, Jenkins, GitHub Actions
Data & AI/ML: PyTorch, Transformers, NLP, Object Detection (YOLOv8), Embeddings & Vector Search
Databases & Messaging: SQL, MongoDB, Redis, Kafka


๐ŸŒ Letโ€™s Connect

Iโ€™m always open to connecting, discussing opportunities, and collaborating on projects.


Pinned Loading

  1. TestStripDX TestStripDX Public

    Forked from dbsymbiosis/TestStripDX

    TestStripDX: An image processing framework for processing and extracting test strip results from a photo

    Jupyter Notebook

  2. ChatFusion-Real-Time-Messaging-with-SmartBot-Integration ChatFusion-Real-Time-Messaging-with-SmartBot-Integration Public

    A real time messaging application, integrated with OpenAI LLM as a chatbot

    TypeScript

  3. Personalized-text-to-image-generator Personalized-text-to-image-generator Public

    Forked from kunaldudhavat/Personalized-text-to-image-generator

    Implementation of the personalized image to text generation framework proposed in the paper: VICO: Plug and Play Visual Condition for Personalized Text-to-Image Generation

    Python

  4. Robot-Guidance Robot-Guidance Public

    Forked from kunaldudhavat/Robot-Guidance

    Designing a crew rescuing bot based on MDPs (Reinforcement Learning) and Machine Learning

    Python

  5. Space-Rescue-Mission Space-Rescue-Mission Public

    Forked from kunaldudhavat/Space-Rescue-Mission

    We are designing different bots which help the space roomba whose crew mates are in danger because of the threat of aliens. We compare the performance of different bots in different scenarios, withโ€ฆ

    Python

  6. Sentence-Level-AI-Generated-Text-Detection Sentence-Level-AI-Generated-Text-Detection Public

    Model for sentence level AI generated text detection. The implementation is based of the paper SeqXGPT which in turn is based on DetectGPT

    Python