VP Vraj Patel AI/ML · Data Science · Full-Stack
OPEN TO WORK
Personnel File — Dossier Open to work
Dossier · Anand, Gujarat, IN · B.Tech IT, Class of 2027

VRAJ
PATEL

AI/ML Engineer · Data Analytics & Data Science · Full-Stack Dev

I build systems that learn and systems that ship — from computer-vision pipelines that restore damaged photographs to civic platforms that route real complaints to real municipal offices. This page is a working record of that: what I've built, how it's built, and what it does.

View Field Work Download RÉsumÉ
8.61Cumulative GPA / 10
07Shipped Projects
02Hackathon Selections
Fig. 1 — How the work movesHover / tap a stage
§ 01 — Profile

Who's behind the work

I'm a B.Tech Information Technology student at MBIT, Anand, Gujarat. I work primarily as an AI/ML Engineer, with a strong interest in data analytics and data science, plus full-stack development to ship what I build — from real-time computer vision pipelines to BI dashboards that turn raw sales data into decisions, to full-stack civic platforms that regular people actually use.

I like the part of the job most people skip: turning a model or a dataset into something someone can actually act on — a dashboard, an interface, a working deployment — not just a notebook or a chart. Fast learner, steady teammate, and I go looking for the harder problem.

DegreeB.Tech, Information Technology
InstitutionMBIT, Anand, Gujarat
Graduating2027
Based inAnand, Gujarat, India
Cumulative Grade Point Average
8.61/10
Semester Performance
§ 02 — Experience

Time on the job

One internship so far — used it to ship a real five-model pipeline, not just watch one being built.

Internship Software Engineer — TenUp Software Services LLP
May 2026 +

Vadodara · 11–22 May 2026. Joined TenUp Software Services LLP as a Software Engineer Intern, working under Vishal Prajapati on applied computer-vision tooling. The brief: design and build an AI image-restoration system end to end — model selection, pipeline architecture, and a usable interface — rather than a single notebook exercise.

Shipped NeuralRestore. Built and delivered a sequential five-model restoration pipeline (EfficientNet-UNet, LaMa, DnCNN, NAFNet, Real-ESRGAN) with a Gradio front end, performance rated as exceeding expectations by the team. Full technical breakdown is in the Field Work section below.

Stack: Python, OpenCV, PyTorch, Gradio, plus the five restoration models chained into one pipeline.

§ 03 — Capabilities

What I build with

Grouped by role rather than buzzword — the tools I reach for at each stage of the pipeline above.

Programming Languages01
PythonSQLHTML5CSS3JavaScript (Basic)
ML & Computer Vision02
PyTorchTensorFlowScikit-learn OpenCVMediaPipe YOLOv11Sentence-Transformers
LLM / RAG & Fine-Tuning03
OllamaChromaDB LoRA Fine-TuningUnsloth Hugging Face PEFTPrompt Engineering GGUF Quantization
Data Analytics & ETL04
PandasNumPyStatsmodels ETL PipelinesData Preprocessing EDA
Data Viz & BI05
TableauPower BI PlotlyMatplotlib Chart.js
Web Frameworks & UI06
FlaskGradioStreamlit
Databases & Version Control07
SQLitePostgreSQL GitGitHub
AI-Assisted Dev & Workflow08
ClaudeCursor AntigravityCopilot Problem SolvingFast Learner
Python
PyTorch
TensorFlow
Scikit-learn
OpenCV
MediaPipe
YOLOv11
Hugging Face
Ollama
ChromaDB
Pandas
NumPy
Python
PyTorch
TensorFlow
Scikit-learn
OpenCV
MediaPipe
YOLOv11
Hugging Face
Ollama
ChromaDB
Pandas
NumPy
Tableau
Power BI
Plotly
Matplotlib
Chart.js
Flask
Gradio
Streamlit
PostgreSQL
SQLite
SQL
Git
GitHub
HTML5
CSS3
JavaScript
Tableau
Power BI
Plotly
Matplotlib
Chart.js
Flask
Gradio
Streamlit
PostgreSQL
SQLite
SQL
Git
GitHub
HTML5
CSS3
JavaScript
§ 04 — Field Work

What I've shipped

Seven projects, three problem domains. Filter by the kind of work, or scroll through all of it.

neuralrestore.app
NeuralRestore interface
01NeuralRestore
01 / Featured Project

NeuralRestore

An advanced AI-powered image restoration pipeline that automatically detects and repairs physical photo damage. Runs a sequential 5-model deep learning architecture — EfficientNet-B4 + UNet, LaMa inpainting, DnCNN denoising, and Real-ESRGAN 4x super-resolution scaling.

PythonPyTorch GradioOpenCV DnCNNLaMa Real-ESRGAN
stocksense.ai/dashboard
StockSense AI dashboard
02StockSense AI
02 / Featured Project

StockSense AI

An intelligent stock analysis and predictive trading insights platform. Combines real-time market data streaming with automated technical analysis (RSI, MACD, Bollinger Bands), interactive candlestick visualisers, and hybrid buy/sell recommendations.

PythonFlask StatsmodelsPlotly PandasXGBoost GARCH
civic.anand.gov.in
Anand Civic System map
03Anand Civic System
03 / Featured Project

Anand Civic System

A smart-city civic issue reporting and municipal management system. Location-aware complaint submission with GPS boundary validation, a live interactive Leaflet GIS map, and Gemini AI image verification before a report ever reaches an officer's desk.

JavaScriptNode.js ExpressLeaflet.js PostgreSQLHTML5/CSS3 Gemini AI
smart-attendance.local
Smart Attendance System
04Smart Attendance
04 / Featured Project

Smart Attendance System

A high-accuracy biometric facial-recognition attendance platform. Uses OpenCV and facial embedding feature extraction for instant live verification, liveness validation to stop photo spoofing, and automated CSV/PDF logging.

PythonOpenCV Scikit-learnFace Recognition SQLiteTkinter
drowsiness-monitor.local
Driver Drowsiness Monitor
05Drowsiness Monitor
05 / Featured Project

Driver Drowsiness Monitor

A real-time, dual-phase computer-vision safety system. MediaPipe Face Mesh drives Eye Aspect Ratio (EAR) calculations and head-pose estimation, paired with a custom YOLOv11 model that catches phone distraction the eye-tracking alone would miss.

PythonMediaPipe YOLOv11OpenCV SciPyPygame
shoppulse.streamlit.app
ShopPulse dashboard
06ShopPulse
06 / Featured Project

ShopPulse

A Power BI-inspired interactive BI dashboard for e-commerce analytics. Executive KPI tracking, RFM customer segmentation, What-If scenario simulation, revenue forecasting, and an AI analytics assistant that answers questions against the live data.

PythonStreamlit Power BIPlotly PandasSQL Scikit-learn
joblens.local / offline-rag
JobLens interface
07JobLens
07 / Featured Project

JobLens

A privacy-first, 100% offline resume-to-job matching system built on a local RAG pipeline. Dense vector similarity search with ChromaDB, synonym-aware skill-gap analysis, and candidate scoring from a fine-tuned local Ollama model — nothing leaves the machine.

PythonOllama ChromaDBSentence-Transformers LoRA / PEFTGradio GGUF Quantization
§ 05 — Competition Log

Under the clock

Two entries so far, both under real time pressure. Expand an entry for the mission, the outcome, and the stack.

36-Hour Sprint CVMU Hackathon 4.0
2026 +

Civic Issue Reporting System. Built a streamlined platform for citizens to report local civic issues — the brief was rapid prototyping and user-centric design, no room for complex overhead.

Delivered. Working prototype shipped before the 36-hour deadline. Validated the core reporting loop and held up as a genuine agile-teamwork exercise under pressure.

Stack: HTML/CSS/JS front end, Python + Flask backend API, SQLite, Leaflet.js for map integration, Claude & Cursor for AI-assisted development.

SIH Qualifier Internal SIH Selection
2025 +

Smart Attendance System. Ideated and built an automated, frictionless attendance solution for the Smart India Hackathon qualifier round, with accuracy, speed, and real-world applicability as the judging bar.

Qualified. Presented the concept to the judging panel, scored well on innovation, and validated the core recognition logic as ready for real deployment.

Stack: Python & OpenCV for vision, MediaPipe & YOLO for detection models, Flask backend, HTML/CSS/JS front end, PostgreSQL.

§ 06 — Contact

Got something
worth building?

I'm open to internships, freelance builds, and hackathon teams. Reach out through whichever channel you actually check.

Emailsp533013@gmail.com → Phone+91 76987 50331 → GitHubvrajptl275 → LinkedInvraj-patel →
Download Full RÉsumÉ

Opens your email client, addressed to sp533013@gmail.com

Ask about Vraj
Portfolio Assistant
Ask about Vraj's work, skills or background
Hey — I'm an intelligent assistant trained on Vraj's portfolio and technical background. Ask me about his skills, projects, work experience, education & CGPA, or how to hire him.