Sagar Adhikari, PhD
Professional Summary
I find the mysteries of the universe to be incredibly fascinating as both an astrophysicist and a data scientist. Working with massive, high-dimensional data from telescopes around the world has honed my expertise in statistical modeling, machine learning, and advanced data visualization. From writing predictive algorithms to developing complex mathematical models, I constantly seek to apply data science techniques to push the boundaries of knowledge and solve broad, real-world problems.
Core Competencies
- Python (Pandas, Scikit-learn)
- Machine Learning & NLP
- Deep Learning (TensorFlow, XGBoost)
- Statistical & Time Series Analysis
- RAG & LLM Integrations
- Data Visualization
Key Projects
My Research Assistant (RAG Pipeline)
Python, NLP, StreamlitAn end-to-end Retrieval-Augmented Generation (RAG) web app built to interactively answer questions related to my peer-reviewed publications. Retrieves relevant responses from dense academic papers on Supermassive Black Holes and AGNs.
Stock Market Forecasting
Pandas, Scikit-learnUtilized multiple statistical and machine learning models to forecast S&P 500 market trends. Compared and evaluated the effectiveness of different models to forecast impending market shifts.
ML Classifications of Fermi-LAT Blazars
XGBoost, Random ForestAnalyzed raw data from the Fermi-LAT telescope for classification of blazar types. Trained Decision Tree, XGBoost, and Random Forest classifiers, achieving over 90% accuracy with GBDT.
Hangman ML Solver
NLP, N-Gram ModelsImplemented an N-Gram model to learn word patterns from a dictionary to solve the Hangman Word Game with ~66% accuracy against the training dictionary.
Selected Publications
- Constraining the PG 1553+113 binary hypothesis: interpreting a new, 22-year period. Analyzed a century-long optical light curve, finding hints of a ~22 yr oscillation and exploring its relationship to supermassive black hole binaries.
- Singular spectrum analysis of Fermi-LAT blazar light curves: A systematic search for periodicity and trends in the time domain. Processed data for 494 sources, extracting oscillatory components from trends and noise.
- Distortions in periodicity analysis of blazars: the impact of flares. Investigated how stochastic flaring events mask or distort true periodic behavior in active galactic nuclei.
Conferences & Achievements
- 2025: $80,000 NASA Fermi-GI research award grant
- 2025: Meeting of Astronomers of South Carolina, Florence, SC, USA
- 2024: 11th International Fermi Symposium, Maryland, USA
- 2023: International Conference on Time Series and Forecasting (ITISE-2023), Gran Canaria, Spain
- 2023: AAS/High Energy Astrophysics Division, Hawaii, USA