Gurv
Fatehabad, Hisar Division, Haryana, India
Experience: 1 years
Open to: Full-Time
Education: Bachelors
Availability: Immediate
Skills: CPP, Docker, Git, Jupyter Notebooks, LLMs/ChatGPT, Matplotlib, MySQL, NumPy, pandas, Python, Python3, PyTorch, scikit-learn, TensorFlow
Previously worked at: Startups, US-based Companies
Assessment Score: 100
- Educational Background
Pursuing a B.Tech. in Information Technology at the Indian Institute of Information Technology, Una, with a strong academic record (CGPA: 8.10). Relevant coursework includes Operating Systems and DBMS, providing a solid foundation in software and systems engineering. - Industry Experience
Over a year of diverse internship experience in machine learning and artificial intelligence across organizations such as mavQ, Bipolar Factory, DataToBiz, and Incruiter. Worked on tasks ranging from improving embeddings in NLP models like BERT and Sentence BERT to deploying complex pipelines for face recognition and classification using cutting-edge technologies like TensorRT and Deepstream. - Technical Expertise
Proficient in Python and C++, with expertise in libraries and frameworks including Pandas, Scikit-learn, PyTorch, TensorFlow, and Docker. Hands-on experience in deploying applications using Flask and FastAPI, alongside proficiency in Generative AI tools like LangChain and LlamaIndex. - Key Projects
Successfully led projects like Autism Disorder Classification, achieving 90% accuracy and 95% sensitivity using Vision Transformers; RAG Question Answering, utilizing Llama2 and ChromaDB for efficient document retrieval and response generation; and an Image Caption Generator, which significantly improved BLEU scores for visual content descriptions through advanced transformer integration. - Achievements and Leadership
Solved over 200 coding problems on platforms like LeetCode and CodeChef, demonstrating analytical and problem-solving skills. Also served as the coordinator for a Table-Tennis sports event at IIIT Una, showcasing organizational and team leadership capabilities. - Performance and Contributions
Consistently improved model performances during internships, such as increasing document classification accuracy to 98.4%, reducing BERT response times by 50%, and cutting paperwork by 70% through summarization automation. Proven ability to fine-tune state-of-the-art models like EfficientNet, YOLOv8, and LLaMA for real-world applications.