PRANSHUL KAUSHIK
Noida, India • +91-914******* • *************@*****.*** •
linkedin.com/in/pranshul-kaushik-8a7795168/
Data Scientist with 5+ years of experience in Data Science and Engineering. Expertise in building production-level applications using Machine Learning, Deep Learning, and GenAI. Skills
● Programming: Python, SQL
● Techniques & Models: Multi-Agent Systems (GCP ADK), LLMs, RAG, BERT, LSTM, LayoutLM, YOLO, Transformer, Named Entity Recognition (NER), Computer Vision
● Libraries & Frameworks: PyTorch, Hugging Face, LangChain, LlamaIndex, Scikit-learn
● Cloud Services: EC2, ECR, ECS, CloudWatch, S3, RDS, Lambda, Textract, Amazon Q Business, Bedrock
● Orchestration & Deployment: PySpark, Apache Airflow, MLflow, Docker, GitHub Actions, CI/CD, Git
● Databases: PostgreSQL, MongoDB, FAISS
PROFESSIONAL EXPERIENCE
Ericsson, Noida, UP - Data Scientist 03/2025 - Present Packet Switched Core Traffic Balancing
Developed a solution to intelligently distribute traffic across multiple Core Nodes, ensuring stable network KPIs and an optimal user experience.
● Engineered a time-series forecasting pipeline using LSTM models to predict utilization patterns across Core Network nodes up to 12 hours in advance, enabling proactive traffic balancing and preventing network congestion before it occurs.
● Multi-Agent AI System: Designed and implemented a chatbot-driven multi-agent system using GCP ADK with orchestrator–worker architecture; built specialized agents for query understanding, SQL generation, anomaly detection, KPI correlation analysis, and automated reporting.
Result: Delivered a proactive, AI-powered traffic management system that significantly reduced network instability incidents and manual intervention overhead, improving Core Network reliability and operational efficiency for end-users. Calance, Gurgaon, Haryana - Data Scientist 04/2023 – 03/2025 AuRA Project
Developed an AI-driven automated research assistant to streamline data collection and analysis.
● Developed a GenAI-powered Knowledge Graph that extracted and stored insights from research papers, enabling advanced searchability and analysis.
● Introduced a dynamic topic refinement pipeline using Retrieval-Augmented Generation (RAG) to improve research accuracy, supported by an intelligent Q&A system that delivered precise and actionable answers from indexed research papers. Result: Produced a sophisticated automated research aide, enhancing efficiency in data collection and analysis, with a notable reduction in research time for end-users. MastersIndia, Noida, UP - Machine Learning Engineer 10/2021 – 04/2023 Information Extraction Project
Spearheaded the development of information extraction problems for streamlined financial transactions. Took the initiative to independently coordinate and manage the project, guiding its development from conception to deployment.
● Developed a robust system for extracting complex information from unstructured documents by fine-tuning LayoutLMv2 and YOLOv5 models, improving the accuracy of Table Extraction and Named Entity Extraction task. Result: Streamlined annotation process, enhancing project efficiency and accuracy for real-world ML implementation. Achieved high extraction accuracy on diverse invoice formats, reducing manual processing time by 70%.
EDUCATION
Shri Ramswaroop College Of Engineering & Management, Lucknow, UP 2017-2021 Bachelor of Technology, Computer Science Engineering