Shreyank Saparia

Data Scientist

Gujarat, India 5+ yrs exp 90 · Outstanding

About

Data Scientist (MSc CS, TU Berlin) with 4+ years in Python and SQL: EDA, statistical analysis, feature engineering and predictive models for time-series forecasting and classification on large datasets. Took ML from data pipelines to production with PyTorch/XGBoost models, MLflow evaluation and monitoring, FastAPI and Kubernetes, plus LLM/RAG solutions that cut manual workload by 40%.

Skills & Expertise (51)

Python Advanced

9.0/10

Years Exp

Generative AI Advanced

8.5/10

Years Exp

Feature Engineering Advanced

8.5/10

Years Exp

Predictive models Advanced

8.5/10

Years Exp

Time-Series Forecasting Advanced

8.5/10

Years Exp

PyTorch Advanced

8.5/10

Years Exp

AWS Advanced

8.0/10

Years Exp

Docker Advanced

8.0/10

Years Exp

Kubernetes Advanced

8.0/10

Years Exp

FastAPI Advanced

8.0/10

Years Exp

SQL Advanced

8.0/10

Years Exp

MLflow Advanced

8.0/10

Years Exp

Next.js NLP Hugging Face LangChain LangGraph OpenAI Google Gemini Claude React Typescript Tailwind CSS Streamlit Agile Development collaboration Vercel GitHub Actions make n8n Pandas EDA Statistical Analysis Classification XGBoost YOLO OpenCV scikit-learn Tesseract Snowflake Keras LLMs NumPy Matplotlib GGplot Plotly Jupyter Notebook TensorFlow CNN RNN Transformers

Work Experience

Full Stack AI/ML Engineer

Upwork Self Employed

Aug 2025 - Present

2D Floor Plan System — real-time object detection and spatial analysis using YOLO + OpenCV (scikit-learn for train-test splitting), extracting measurements and structured data from floor plan images. Computer Vision Benchmarking: fine-tuned YOLO models from 45% to 92% accuracy; compared YOLOv8, Detectron2 and MiDaS using model evaluation metrics (precision/recall, mAP). Engineered a FastAPI backend with an MCP server for product sync, semantic search, and LLM automation — cut catalog sync from 4 hours to 15 minutes across 20K+ SKUs. LuminaFix — computer vision model for style transfer with region-aware processing, deployed as a live web app.

AI/ML Engineer (Contract)

Vodafone

Feb 2025 - Jul 2025

Worked with stakeholders on process mapping and prototyping to automate employee IT issue resolution with a LangChain + OpenAI agent, reducing manual ticket handling by 40%. Developed OCR-powered system using Tesseract and Python to automate processing of employee operational forms, reducing HR processing time by 65% and handling 200+ submissions daily.

Machine Learning Engineer

Fraunhofer Fokus - Applied Research Institute

Jun 2021 - Dec 2024

Built time-series forecasting/classification models (PyTorch) for 5G resource management: feature engineering on large 5G datasets via Snowflake pipelines; model evaluation and monitoring with MLflow; 80% accuracy. Shipped 5G core software to AWS(ECS) cloud infrastructure setup and enabled single click deployment achieving 50% optimization. Deployed Semaphore using containerization (Kubernetes & Docker) by helm charts in on premise environments.

Data Analysis

Byjus

Apr 2018 - Dec 2018

Developed Tableau dashboards for data visualization and identifying key learning preferences to enhance content strategies. Built and optimized data pipelines to clean and preprocess student data into structured insights on learning behavior.