CV

Roshan Kumar — experience, education, projects and skills.

Contact Information

Name Roshan Kumar
Professional Title Data Engineer / Business Analyst

Professional Summary

Data engineer building PySpark and Databricks pipelines for supply-chain analytics. Background in state estimation and robotics from IIT Guwahati. Interested in learning for control.

Experience

  • 2026 - present

    Gurugram, India

    Senior Business Analyst
    SCA Technologies
    Own the data pipelines and reporting layer behind supply-chain analytics for large restaurant and retail networks. Promoted twice in two years — Trainee Analyst (Jun 2024), Business Analyst (Jun 2025), Senior Business Analyst (Jul 2026).
    • Built and own end-to-end PySpark/Databricks pipelines populating the core fact tables — distribution-centre performance, sales, and vendor performance.
    • Designed the fact-table architecture behind the analytics layer.
    • Built and maintain a custom COE REST API.
    • Own the production Tableau reports. Debug and extend reports including DC Performance Summary, COOP Summary Monthly, and the Systemwide RFE Trend report.
  • 2023 - 2023

    Remote

    Data Analyst Intern
    TransOrg Analytics
    • Analysed client data in the automotive sector to find pricing trends and support pricing strategy.
    • Built an interactive Tableau dashboard on SQL-prepared data and trained XGBoost and gradient-boosting models to predict car prices.
  • 2023 - 2023

    Guwahati, India

    Research Intern, under Dr. Budhaditya Hazra
    IIT Guwahati
    Function estimation on Riemannian manifolds.
    • Explored Riemannian geometry and Itô calculus to capture nonlinear relationships using geodesics and tensors.
    • Estimated parameters of Gaussian and log-normal densities on curved spaces with MLE and Bayesian models.

Education

  • 2020 - 2024

    Guwahati, India

    B.Tech
    Indian Institute of Technology Guwahati
    Civil Engineering
    • Thesis (supervised by Prof. Budhaditya Hazra) — Parameter and state estimation of vibrating structures equipped with geometrically consistent tuned mass dampers, using Kalman filters on Riemannian manifolds.

Projects

  • 2026 - 2026
    Smart Factory Scheduling — an OpenEnv RL environment

    OpenEnv-compliant reinforcement learning environment for industrial job scheduling: assign jobs to machines under stochastic breakdowns and hard deadlines. Packaged as a Docker-based Hugging Face Space.

    • Python, reinforcement learning, OpenEnv, Docker
  • 2025 - 2025
    SlideGen — MCP toolchain

    Three Model Context Protocol servers powering a document-to-slides workflow: a webpage scraper/summariser, a PDF analyser, and a PowerPoint generator.

    • Python, MCP, uv
  • 2023 - present
    Chess Engine

    Chess engine written from scratch in Python (2023), revisited in 2026 to measure and improve it. Raised its strength from about 1060 to about 1650 on Stockfish’s rating scale, measured in 100-game matches per change.

    • Validated move generation against published perft counts and python-chess: fixed six rules bugs, then zero mismatches across 573,000 positions.
    • Added a UCI interface and a reproducible match harness against Stockfish at fixed ratings, with confidence intervals and every game saved.
    • Added and measured each change separately: quiescence search with move ordering (+205), a transposition table (no measurable gain), piece-square evaluation (+335).
    • Python, alpha-beta search, UCI, pytest, GitHub Actions
  • 2025 - 2025
    Context-Aware Document Q&A

    Retrieval-augmented generation over PDFs and web content — sentence-transformers embeddings, FAISS vector store, Gemini for answer generation, Streamlit front end.

    • Python, RAG, FAISS, sentence-transformers, Streamlit

Awards

  • 2023
    Gold Medal, Inter-IIT Tech Meet 11.0
    DevRev

    DevRev’s “Expert Answers in a Flash” at Inter-IIT Tech Meet 11.0. Built a knowledge-base retrieval and QA pipeline with GenQ-trained retrievers, NER re-ranking, and ONNX/quantised models that brought latency to about 250 ms.

  • 2022
    Bronze Medal, Inter-IIT Tech Meet 10.0
    Bosch

    Age and gender detection from low-quality CCTV footage (YOLOv5, ByteTrack, SwinIR, Swin Transformer with a CORAL layer): 94.4% accuracy, age MAE of 5.25.

  • 2022
    IEDC Funding — AntahSagari
    Innovation and Entrepreneurship Development Centre, IIT Guwahati

    Finalist for Government of India funding to prototype and patent an underwater vehicle.

  • 2022
    International Rover Design Challenge

    Qualified in the top 16 teams worldwide.

  • 2022
    Best Project, Techevince 8.0

    Best Project in the Judges’ Choice and Overall categories, with Yuvaan.

Skills

Data Engineering: PySpark, Databricks, SQL, Delta Lake, ETL design, fact-table modelling
Analytics & Reporting: Tableau, REST API design, data quality investigation
Machine Learning: PyTorch, reinforcement learning, RAG, sentence-transformers, FAISS
Estimation & Control: Kalman filtering, unscented Kalman filters on manifolds, MATLAB
Tooling: Python, Git, Docker, Streamlit

Languages

English : Fluent
German : Beginner
French : Beginner