Pratham Dabhane

pratham.dabhane.2503@gmail.comMumbai, India

Summary

An AI & Data Science graduate with hands-on experience building backend APIs and data pipelines using Python and FastAPI. Has worked with vector databases, and integrated LLM APIs in project-based applications. Familiar with cloud-hosted services like Supabase and Render through personal and hackathon projects. Comfortable working with REST APIs, structured data, and deploying small-scale systems end-to-end.

Education

Terna Engineering College, B.E., Artificial Intelligence and Data Science
2022 - 2026

Navi Mumbai, India, Grade: 7.52

Skills

Python, SQL, Java, Machine Learning, Deep Learning, Generative AI, LLMs, RAG, Prompt Engineering, NLP, Computer Vision, Scikit-learn, Pandas, NumPy, TensorFlow, LangChain, FastAPI, Flask, Streamlit, REST APIs, Qdrant, Pinecone, Supabase, PostgreSQL, MySQL, MongoDB, Docker, Git, GitHub, Linux, MLOps, MLflow, Data Analysis, K-Means, PCA, SVM, CNN, DenseNet, A* Pathfinding

Projects

Terrain Simulation & Mission Planning github.com/Pratham-Dabhane/Terrain-Sim-and-Mission

Procedural terrain generation and mission-planning system using fBM, domain warping, erosion simulation, DEM calibration, biome classification, and A* pathfinding.

AlignCV – Resume Fit Scorer github.com/Pratham-Dabhane/AlignCV

AI-powered resume–job matching system using embeddings, Qdrant, Groq LLMs, FastAPI, Streamlit, and Supabase to analyze and score resume compatibility.

Lectra – Document Search & QA github.com/Pratham-Dabhane/lectra

RAG-based document question-answering system that processes documents, generates embeddings, performs vector search, and uses an LLM to provide contextual answers.

Publications

Generative AI for Terrain Simulation and Scenario Planning dx.doi.org/10.2139/ssrn.6539923

International Journal of Progressive Research in Engineering Management and Science (IJPREMS), Vol. 06, Issue 05, May 2026

Research paper presenting a multi-stage procedural terrain synthesis pipeline that combines DEM calibration, fractal Brownian motion (fBM), domain warping, hydraulic and thermal erosion, biome classification, and A*-based mission path planning. The system was validated against real-world SRTM DEM data from the Grand Canyon region and demonstrated improved geomorphometric alignment while maintaining computational efficiency.

Languages

  • English, Fluent