Sahil Yadav: Civil Engineer & Data Scientist
Uploaded by
Sahil YadavSahil Yadav: Civil Engineer & Data Scientist
Uploaded by
Sahil YadavCommon questions
Powered by AISahil Yadav managed security operations for IIT Roorkee’s annual cultural festival, THOMSO 2022, ensuring smooth event execution. This role involved coordinating complex logistic and safety processes, which enhanced his problem-solving abilities by requiring quick decision-making and strategic planning under pressure .
Sahil Yadav developed a RAG application that processes user queries to deliver precise, context-sensitive responses using LLM. He enhanced query latency by 15% and retrieval accuracy by 32% by integrating Langchain-Pathway. He also introduced techniques such as zero-shot inference, prompt templating, and entity disambiguation .
Sahil Yadav innovated data analytics techniques to enhance digital presence, which led to an improvement in the lead generation rate by 1-2%. Additionally, he implemented a responsive UI using HTML5, SCSS, and JavaScript, which increased user engagement by 30% .
Sahil Yadav unified HPC and PyTorch’s DistributedDataParallel to parallelize training across multiple GPU nodes, which optimized the performance and accelerated the training time of the self-driving car project. This approach helped achieve an accuracy of up to 85% and reduced the system latency by 8% while increasing accuracy by 7% when trained on YOLOV9 using HPC .
During his internship, Sahil Yadav incorporated diverse environmental datasets, including river flow and meteorological variables, to design predictive models for rainfall forecasting. He employed LSTM networks, which improved the accuracy of the predictions by 18%. He also deployed various models including Regression, Random Forest, AdaBoost, Gradient Boost, XGBoost, and RNN .
Sahil Yadav incorporated OpenCV and TensorFlow/Keras libraries to resize, normalize, and augment the dataset by 15-20%. He introduced various optimizers like RMSProp and Momentum SGD, which improved the overall accuracy of the system by 15% and engineered CNN architectures such as Resnet and VGG, achieving a classification accuracy of 92% across datasets .
Sahil Yadav engineered a high-performing LLM app through advanced prompt engineering and hyperparameter tuning. He developed a high-precision fine-tuned LLM (scoring between 0.7-0.8) for domain-specific question answering. His innovations resulted in a 20% increase in contextual response accuracy of the fine-tuned model .
Sahil Yadav engaged in over 15 competitive events and team-building activities over two years within the IIT Roorkee Taekwondo Club. His involvement fostered a strong sense of camaraderie among club members through skill collaboration and participation .
As an executive member of the Himalayan Explorer’s Club, Sahil Yadav organized an adventure expedition for over 200 participants, emphasizing safety and logistical coordination. He also promoted outdoor exploration and environmental awareness among more than 10 club members .
Sahil Yadav formulated a Transformer-based encoder and StyleGAN to generate high-resolution images from text. He established pipelines for preprocessing, training, and validating models on a large-scale dataset of approximately 1 GB. By fine-tuning the transformer for textual interpretation, he refined semantic comprehension by 25% .