Work
A collection of projects, research papers, open-source contributions, and recognitions in AI/ML engineering and research.
Built an LLM-powered dungeon master capable of generating dynamic narratives and branching storylines.
•Implemented turn-based combat logic, state persistence, and procedural world generation.
•Designed a Streamlit-based interface enabling real-time interaction with the AI system.
Contributed to a large-scale open-source repository recognized by GSSoC, Hacktoberfest, and SSOC.
•Worked across machine learning, automation, and cybersecurity modules.
•Reviewed pull requests, improved documentation, and mentored new contributors.
Designed a scalable RAG pipeline combining vector databases and large language models.
•Implemented efficient document embedding, retrieval, and prompt construction strategies.
•Reduced hallucinations by grounding responses in retrieved contextual data.
Developed CNN-based models for tumor detection from MRI and CT imaging data.
•Applied preprocessing, augmentation, and architectural tuning to improve robustness.
•Deployed inference pipelines using Flask and Docker for scalable access.
Built time-series forecasting models using LSTM, ARIMA, and Prophet.
•Modeled disease progression using EHR and clinical time-series data.
•Focused on interpretability and early risk identification.
Implemented text-to-image generation using Stable Diffusion and CLIP guidance.
•Tuned prompts and latent-space parameters for domain-specific image generation.
•Deployed the system via a Gradio interface for interactive use.
Digital Device for Vedic Knowledge Analysis
research
Indian Design Patent No. 447520-001 (2025) — Co-Inventor
•Designed an AI/NLP-based platform for digitizing and analyzing ancient Vedic texts.
•Focused on automated text processing, semantic analysis, and knowledge extraction.
•Filed and granted under the Indian Design Patent system.
Springer LNNS (2025) — Co-Author
•Analyzed EEG signals using machine learning for neurological disorder detection.
•Compared multiple classifiers including Gradient Boosting, HMM, and AdaBoost.
•Achieved improved sensitivity over baseline methods.
AIP Conference Proceedings (2025) — Co-Author
•Applied predictive modeling to estimate protein–ligand binding affinities.
•Utilized the BELKA dataset with millions of molecular samples.
•Aimed at accelerating early-stage drug discovery pipelines.
Springer Energy (2025) — Co-Author
•Studied piezoelectric materials for sustainable energy harvesting.
•Analyzed material properties and real-world applications.
•Proposed hybrid material use for low-power IoT systems.
Cyber Security Grand Challenge 2.0 — Idea Stage Winner
achievement
Government of India (MyGov) — 2025
•Ranked among top submissions out of 1,000+ national participants.
•Proposed a quantum-resistant cybersecurity framework.
•Presented the solution to government and industry experts.
NeurIPS 2024 BELKA Competition — Bronze Medalist
achievement
Kaggle — Top 6% (Ranked 118th out of 1,946 teams)
•Built scalable ML models for molecular binding prediction.
•Applied feature engineering and deep learning techniques.
Certifications
AI Agents Fundamentals
Hugging Face
2025
Career Essentials in Generative AI
Microsoft & LinkedIn
2024
OpenCV Bootcamp
OpenCV University
2024
Machine Learning with Python
freeCodeCamp
2024
Data Science using Python
IBM
2024
Introduction to AI with Python
edX
2024
Introduction to Responsible AI
United Latino Students Association
2024