I am Nishad Nakhwa,
a Graduate Research
Assistant & Data Analyst
About
I'm a Graduate Research Assistant and Data Analyst passionate about using data to drive intelligent solutions. With a Master's degree in Data Science from Florida State University, I specialize in turning complex datasets into clear, actionable insights through predictive modeling, data visualization, and scalable analytics pipelines. My work also spans natural language processing, where I fine-tune large language models using LoRA and PEFT with tools like PyTorch and Hugging Face. I haved worked on mordern AI techniques like RAG while building a chatbot. I thrive at the intersection of analytics and AI—solving real-world problems with a blend of statistical thinking and machine learning.
Expertise
Turning ideas into impact with these skills:
Programming Languages & Frameworks
R, Python, Pytorch, Transformers, TensorFlow.
Data Analysis Tools
NumPy, Pandas, Seaborn, Matplotlib, SciPy, Scikit-learn.
Database Management
SQL (MySQL, PostgreSQL)
Visualization Software
Microsoft Power BI, Tableau, Microsoft Excel, Google Looker
Cloud Technologies
AWS (EC2, S3, RDS), Azure, Snowflake.
Data Science & Analytics
Data Analytics, Data Visualization, Data Mining, Data Modeling, Machine Learning, Predictive Modeling.
Experience
Florida State University - Department of Scientific Computing
Graduate Research Assistant
April 2025 - Current
- Building a RAG chatbot for the Department of Scientific Computing website using Selenium for web scraping, converting content into JSONL, and implementing document chunking with vector embeddings.
- Implementing search over a vector store and integrated the gpt-oss model using Ollama and LangChain, enabling accurate, department-specific responses to user queries.
- Fine-tune LLMs on datasets with over 50,000 examples using LoRA and PEFT, improving task accuracy by 10-15% on models such as DistilGPT2 through parameter-efficient techniques and controlled experiments.
- Build and optimize fine-tuning pipelines with Hugging Face Transformers and PyTorch, achieving 85%+ F1 scores acrosssentiment classification, summarization, and Q&A, while reducing training time by 35% using mixed-precision and batch tuning.
- Collaborated with faculty and peers across disciplines to interpret data insights, documented findings, and presented results viatechnical reports and visual dashboards.
Bartech Data Systems Pvt. Ltd.
Data Analyst
June 2022 - August 2023
- Designed and deployed scalable data models to process RFID/barcode data, improving logistics traceability and reducingtracking delays by 30% across manufacturing and pharma clients.
- Developed business intelligence dashboards in Power BI to track KPIs like machine utilization and delivery timelines, reducingreporting latency by 40% through automated refresh scheduling and optimized queries.
- Managed high-volume datasets in Snowflake and AWS S3, and automated ETL workflows to support real-time analytics, improving visibility into stock movement and delivery performance by 15%.
- Collaborated with cross-functional teams using Agile methodologies and Jira to align data solutions with evolving clientrequirements, accelerating project delivery by 20% and boosting platform adoption during client onboarding.
Education
Florida State University
Master of Science in Data Science - Computer Science
May 2025
Relevant Coursework includes Machine Learning, Data Mining, Regression Analysis, Advanced Data Science, Data Ethics, Mathematics for Data Sciecne, Computer Security in Data Science.
University Of Mumbai
Bachelor of Engineering in Information Technology
May 2022
Relavant Coursework includes Data Mining & Business Intelligence, Artificial Intelligence, Big Data Analytics, Cloud Computing, Data Structures & Analysis, Database Management System.
Recent Works
Here are some of my favorite projects I have done lately. Feel free to check them out.
Fine-Tuning LLM using LORA
Adapted DistilGPT-2 on the Tiny Stories dataset with PEFT and LoRA for efficient fine-tuning. The model now creates engaging, context-aware short stories from simple user prompts.
Data-Driven Fitness Tracker
Built a personalized fitness companion that tracks BMI, weight goals, diet, and exercise plans. Features an exercise log with intensity graphs to turn progress into clear, actionable insights.
Body Perforance Analysis
Explored Korean Olympic athlete data to investigate factors influencing grip force. Applied multiple regression, normality tests, and visual analysis in R to uncover key patterns.
Personality Prediction using Social Media
Uncover personality traits through tweets and text analysis using MBTI types. A sleek Flask app powered by TF-IDF and cosine similarity delivers instant personality insights.
RAG Chatbot
Designed a smart departmental assistant by scraping the department's website data with Selenium. Using RAG with LangChain, and LLaMA 3.1, it delivers accurate, context-aware answers to student queries.
Movie Recommendation System
Developed a personalized movie recommender that suggests films based on user input. Utilizes content-based filtering techniques to deliver relevant and engaging movie choices.
Get In Touch
I'd love to hear from you! Whether you have a question or just want to chat about AI, data science, or tech—feel free to reach out.