Shivamshu Roy
Data Scientist & Machine Learning Engineer
MS in Data Science @ SUNY Buffalo (Dec '27) • Published IEEE Researcher
Shivamshu Roy
Proof of Work & Technical Impact
First-principles machine learning, published research, and enterprise data systems.
Published first-author researcher in financial time-series forecasting (CVMI-2023).
Full-Time Data Scientist (Nxtwave) • Data Science Intern (TCS) • Computer Vision Intern (NIC).
Stockfish 16 NNUE neural evaluations compiled to WebAssembly running 24+ plies deep.
B.Tech CS (Data Science) • M.S. Data Science at State University of New York at Buffalo.
Featured Projects
Production AI systems, chess neural evaluations, and deep predictive models
CheckmateLab
Client-Side WebAssembly Neural Engine (Stockfish 16 NNUE)
High-performance chess analysis platform executing Stockfish 16 NNUE client-side via WebAssembly (WASM). Features multi-threaded Web Workers utilizing SharedArrayBuffer to isolate 24-ply alpha-beta minimax search from the DOM, integrated with Firebase OAuth and real-time player ELO rating progression ($0 cloud compute).
ClearHire AI
Recruiter Inbox LLM Parser & Predictive Response Velocity Engine
Production GenAI pipeline integrating Pydantic-enforced structured schemas with a statistical response hazard model over a containerized FastAPI backend. Hardened against LLM schema drift via automated PyTest regression evals (PR-AUC = 0.94).
Green Grid AI
EV Demand Forecasting & Constrained Facility Location Optimization
Dual-stage municipal EV infrastructure engine: couples an XGBoost demand regression model (R² = 0.89, MAPE = 6.2% across 17,280 hourly records) trained on temporal lag features & diurnal encodings with a Constrained Spatial Clustering & Facility Location algorithm (P-Median / Voronoi) balancing grid capacity and traffic density.
Improving Stock Market Forecasting Using Advanced Time Series Analysis Techniques
Presented at the CVMI-2023 conference and published in IEEE Xplore. Formulated a Nelder-Mead simplex optimization algorithm to dynamically calibrate smoothing constants (α, β, γ) and autoregressive coefficients across non-stationary regimes, achieving an ~18% reduction in out-of-sample forecast error (RMSE/MAE) over standard ARIMA and Holt-Winters baselines.
Want to build something together?
Have an ambitious machine learning idea, hackathon team in need of an ML engineer, or an open Summer 2027 internship role? I'm always open to collaborating with fellow builders, researchers, and technical recruiters on cutting-edge data science and AI systems.
Writing & Technical Insights
Thoughts on AI engineering, geospatial intelligence, learning habits, and building software
There is an odd sort of guilt that accompanies starting a Master’s degree
How a daily 15-minute GitHub habit changed the way I learn, stay accountable, and navigate the transition into graduate data science studies at SUNY Buffalo.
I Got Tired of Paying for Chess Subscriptions, So I Built My Own Browser Engine
A breakdown of CheckmateLab: integrating Stockfish 16 NNUE, WebAssembly, live ELO tracking, and real-time position evaluations right in the browser.
Unravelling the Impact of Noisy Data on Exploratory Data Analysis
Why real-world data is seldom clean: an analytical dive into identifying corrupted signals, handling outliers, and preserving statistical validity during data exploration.
GeoZapp — EVC India: Tackling Electric Vehicle Range Anxiety
How we engineered an intelligent geospatial solution to discover, map, and navigate EV charging stations across India to support sustainable clean mobility.
Experience & Education
Industry analytics, data engineering, and academic background
Professional Experience
Data Scientist
Full-TimeNxtwave Disruptive Technologies
- Architected end-to-end ETL data pipelines using SQL, Google BigQuery, and Python, automating cross-functional reporting across operational teams and reducing manual data processing latency by 40%.
- Designed interactive enterprise analytics dashboards querying 500K+ transactional student records across 4 major business departments, cutting executive report turnaround from 2 days to near real-time.
- Built automated data validation and anomaly detection workflows using Google Apps Script and REST APIs, eliminating multi-team discrepancies and auditing pipeline integrity.
Data Science Intern
InternshipTata Consultancy Services (TCS)
- Developed Python-based automated exploratory data analysis (EDA) and multivariate statistical modeling pipelines using Pandas, NumPy, and Scikit-Learn, decreasing data hygiene and feature preparation time by 30%.
- Formulated hypothesis testing procedures and engineered real-time KPI tracking dashboards in Tableau for multi-disciplinary engineering teams, improving cross-functional reporting velocity by ~25%.
Python Trainee (Intern)
InternshipNational Informatics Center (NIC)
- Developed automated computer vision pipelines with OpenCV to detect, localize, and redact biometric fingerprints in digitized records at 98%+ precision.
- Built containerized image processing microservice APIs using Flask, enabling high-throughput automated document redaction.
Data Science Intern
InternshipTriyas Tech Solutions
- Engineered scalable data preprocessing pipelines and automated feature extraction workflows using Python and SQL across 50k+ sample records.
- Trained, validated, and benchmarked supervised machine learning models for classification and regression tasks, creating interactive analytical dashboards.
Education & Credentials
M.S. in Engineering Science (Data Science)
State University of New York at Buffalo · Buffalo, NY
Incoming Fall 2026 · Eligible for Summer 2027 Internships
B.Tech in Computer Science (Data Science)
Bennett University · Greater Noida, India
Graduated with 8.93 / 10.0 CGPA
Technical Arsenal
Core technologies, machine learning frameworks, and analytics stacks
Let's Connect & Collaborate
Available for Summer 2027 Data Science & ML internships, research discussions, and interesting technical projects.