Riddhi Gohil

Part of the Aurora Serverless team at AWS.

Google Developer Group on Campus Lead

In my role as a Software Development Engineer at Onix, I devised an AI chatbot solution for Google Chatspaces, streamlining the employee matching process and reducing project allocation time by 50%. Additionally, I deployed a serverless DBT application on GCP Cloud Run, automating workflows and improving pipeline efficiency by 80%. I also developed co-agentic workflows using LangChain and LLMs to automate code generation, testing, and deployment, increasing development speed by 20% and reducing manual testing time by 15%.

• Empowering and supporting women-identifying students in tech through knowledge expansion, academic and career growth, and a strong, supportive network. • Women In Tech platform, enriched by community events and insights from Northeastern alumni, fosters thriving members academically and professionally in an inclusive and empowering environment, utilizing on-campus resources for holistic support.

• Developed and implemented the overall design and marketing strategy aligned with the business objectives. • Design and market materials as and when needed.

Teaching Assistant for CS 5010: Programming Design Paradigm (using Java) • Guided students through course materials and helped them understand crucial concepts. • Graded assignments of all students in a timely and fair manner. • Provided guidance and support to students during lab sessions.

• Architected a high-performance ETL framework using Python, Spark, Kafka, and Hadoop to process 1TB of real-time data from RSS feeds into HDFS, orchestrated with Airflow to complete in under 3 minutes. • Migrated the entire rumor pipeline codebase from Python 2.7 to Python 3.9, consisting of 2000+ lines of code. Addressed 50+ compatibility issues and reduced dependencies by 15%, resulting in a more efficient and maintainable codebase. • Fine-tuned SVM models for rumor detection application, achieving a prediction accuracy of 92%. • Streamlined an automated KYC verification system utilizing YOLO object detection and OCR technologies to process 100,000+ daily documents, reducing manual data cleaning by 40% and improving extraction accuracy by 25%. • Formulated an automated Video-to-Text processing pipeline achieving 95% transcription accuracy, integrating CI/CD practices to streamline video collection, MP3 conversion, and speech extraction workflows.

• Spearheaded the enhancement of the file redistribution microservice, optimizing file distribution across servers based on load and vacancy, leading to a 150% improvement in file processing time and operation efficiency, and boosting server performance. • Worked on selenium for web scraping from dynamic websites, processing it, and dumping it into PostgreSQL for structured storage, after normalization and outlier detection, to enhance data accuracy and reliability. • Extracting critical information from scanned documents by utilizing Object Detection Models and Computer Vision libraries, such as PyTesseract OCR(Optical Character Recognition) and OpenCV.