Nikhil Naik

Driving innovation in healthcare with cutting-edge AI solutions that redefine patient interaction.

• Developed a Custom LLM System: Leading the end-to-end development of a healthcare-specific Large Language Model system, focusing on providing accurate, reliable, and context-sensitive responses to medical queries.
• Data Collection and Curation: Gathered and cleaned public chat-format data, utilizing T5 and BERT models to create a masked prediction task for correcting grammatical and other errors, ensuring high-quality datasets for LLM training.
• Fine-tuned Advanced LLMs: Fine-tuned and instruction-tuned multiple advanced language models, including Mistral MOE, Llama3, Llama3.1, and Gemma 2, optimizing them for specialized healthcare use cases.
• Implemented PEFT Techniques: Utilized Parameter-Efficient Fine-Tuning (PEFT) techniques such as LORA and Q-LORA to achieve faster and more efficient model training. Integrated accelerator tools like UnSloth and FlashAttention to enhance the performance of these fine-tuned models.
• Developed Inference Web Application: Designed and deployed an inference web application using Triton Inference Server, TensorRT, and vLLM. The frontend was built with Gradio, creating a user-friendly interface for testing and real-time interaction with the LLM.
• RAG-Based Applications: Currently working on Retrieval-Augmented Generation (RAG) based applications, enabling the system to retrieve and process information from proprietary pharmaceutical and insurance datasets, delivering accurate and contextually relevant responses.
• Exploring SOTA Techniques: Testing and integrating state-of-the-art techniques such as graph RAG and memory tuning to further enhance model performance and ensure the system can handle complex queries with high accuracy and reliability.

Engineered robust security solutions and led transformative projects that helped secure multi-million dollar contracts.

• Data Curation and Model Development: Utilized Python and Scikit-learn to collect, clean, and curate data from various firewalls, forming the foundation for a malicious email detection system. Achieved a 96% recall rate, significantly improving system efficiency by 40%.
• Natural Language Processing (NLP) and Chatbot Development: Engineered a RASA-based Technical Assistant Chatbot to streamline support operations. Developed a custom document parser to interpret complex technical documentation, reducing ticket escalations to engineering teams by 30%.
• Leadership in Cloud Infrastructure and Networking: Led a team in the design and development of Cloud-managed VPN and SD-WAN projects, employing open-source Linux libraries and MySQL. Provided architectural direction and implemented critical features, contributing to a successful $6 million contract.
• Frontend Development and Data Visualization: Modernized firewall UI using ExtJS, Pygal, and Chartjs, creating an intuitive and responsive interface. Enhanced user experience and facilitated real-time data visualization for monitoring and management.
• Project Management and Client Collaboration: Spearheaded the deployment of over 3,000 firewalls in collaboration with clients, ensuring seamless integration with cloud-based management systems. Managed end-to-end project execution, maintaining clear communication with stakeholders.

Graduated at the top of my class with cutting-edge research in Computer Vision and NLP, represented Stevens at an international AI conference.

• Academic Excellence: Graduated with a 3.96 CGPA, achieving on of the top ranks in the class.
• Graduate Research in Computer Vision: Selected by renowned researcher Prof. Enrique Dunn for a high-level academic research project. Collaborated with PhD candidates to tackle challenges in optical flow for motion and depth estimation. Developed advanced software using OpenCV & Open3D for 3D reconstruction mapping and evaluated its performance against COLMAP.
• Graduate Research in NLP and Language Models: Chosen to represent Stevens at WMT 2024 under the guidance of Prof. Jia Xu. Developed an innovative multilingual translation system, improving BLEU scores to 59.8 through iterative back-translation fine-tuning of NLLB and MT-5 models. Co-authoring a research paper detailing these methodologies and results.
• Core Coursework: Completed advanced courses in Natural Language Processing, Deep Learning, Fundamentals of Machine Learning, Statistical Machine Learning, and Knowledge Discovery with Data Mining, along many others.
• Project Experience: Successfully completed multiple projects both individually and in team settings, demonstrating proficiency in collaboration, leadership, and technical execution.

Distinguished graduate with hands-on innovation in ASL translation and a track record of impactful project experience and certifications.

• Graduated with Distinction: Earned a bachelor's degree in Electronics Engineering, graduating with distinction and a solid academic foundation.
• Sign Language Interpreter Glove Project:Developed an innovative hand-worn glove to translate American Sign Language (ASL) gestures into English, utilizing sensors, microcontrollers, and Bluetooth technology. This final year project, which involved proprietary logic for hand-gesture detection, was showcased at national competitions and published in the 2nd International Conference on Advanced Trends in Engineering (ICATE-2k18).
• Certifications and Continued Learning:Completed multiple certifications including IIT Roorkee's Applied Data Science course (2021), Fundamentals of Data Mining by Pirple.com (2021), Python programming by Prognoz Technologies (2018), and C programming by Seed Infotech (2017). Also gained expertise in Raspberry Pi-based system design and development.
• Active Role in Technical and Social Organizations:Actively involved as a member of IETE DMCE, organizing technical workshops and events. Appointed as a Member of the Junior Red Cross since 2019, managing social service initiatives like orphanage visits, charity drives, and arts competitions.
• Project Experience: Successfully completed a range of projects, demonstrating strong problem-solving skills, technical proficiency, and the ability to work effectively both individually and in collaborative team environments.