Aug 2024 – Feb 2025 · 6 mos · Gurugram, Haryana, India · On-site

• Conducted in-depth research on telecom scams in India, identifying prevalent SMS/text fraud patterns. Developed robust spam detection models using traditional ML algorithms like Decision Trees, Random Forest, and XGBoost, achieving accuracy up to 91%, with comprehensive documentation.
• Performed Parameter-Efficient Fine-Tuning (PEFT) using QLoRA to adapt large language models for telecom applications, ensuring high precision and reliability in industry-specific contexts.
• Developed and deployed generative AI systems for advanced text analysis using Hugging Face, Ollama, and LangChain, integrated through FastAPI for scalable, real-time fraud detection. Implemented load balancing to optimize system performance and minimize latency.

May 2024 – Jul 2024 · 2 mos · New Delhi, Delhi, India · On-site

• Analyzed user activity on the Sakhi platform using SQL and Excel, identifying peak usage hours and evaluating departmental performance. Leveraged these insights to implement data-driven changes, leading to a 34% increase in user engagement.
• Performed sentiment analysis using NLP techniques to identify and mitigate harmful or inappropriate content. Implemented topic modeling and content classification algorithms to analyze and categorize the types of content posted, ensuring compliance and enhancing platform integrity, user satisfaction, and trustworthiness.
• Developed a content summarization tool using Hugging Face Transformers and LangChain, enabling efficient text summarization and information retrieval. This tool allowed users to quickly access concise and relevant insights from multiple updates, significantly improving time efficiency.

Oct 2023 – Jan 2024 · 3 mos · Mumbai, Maharashtra, India · Remote

• Engineered advanced time series models including ARIMA and SARIMA for precise telecom network forecasting, enabling accurate predictive scheduling of energy cells. The pipeline also included data preprocessing, cleaning, and transformation, ensuring robust model performance.
• Reduced energy consumption by 8% and optimized cell utilization using ML-based load balancing, ensuring high service quality and meeting key KPIs. Deployed solutions with Streamlit for visualization and FastAPI for scalable backend support.
• Performed data analytics to extract key insights and presented simplified findings to stakeholders, supporting their decision-making process. Collaborated with senior ML engineers to refine the analysis and enhance network operations and efficiency.

Jul 2023 – Aug 2023 · 1 mo · New Delhi, Delhi, India · Remote

• Designed recommendation systems for a social media use-case, advising users on what to post and recommending similar posts based on their interactions, resulting in a 23% boost in user engagement.
• Enhanced content safety by applying NLP techniques and leveraging Hugging Face Transformers to detect and mitigate harmful content. Implemented sentiment analysis and AI-driven content moderation, along with video quality enhancement through ESRGAN for upscaling.
• Developed a nudity-detection filter using open-source models like NudeNet, improving content safety and creating a more user-friendly experience, while optimizing video delivery for a seamless user experience.

Dec 2019 – Dec 2019 · 0 mo · New Delhi, Delhi, India · On-site

• Selected among the top 1% of students across India for the prestigious INSPIRE Internship Programme by the Department of Science and Technology, Government of India.
• Gained invaluable insights through interactions with industry leaders in Science and Technology, learning from their experiences and innovations.
• Conducted hands-on experiments across diverse scientific disciplines, including Botany, Zoology, Physics, Chemistry, and Biochemistry, fostering a multidisciplinary understanding of science.