Shail K Patel

Founding ML Engineer at Stealth Startup building full-stack AI systems from the ground up. I turn product vision into working architecture . Let's collaborate and build a scalable prototype.

Hire Me Resume

Work Experience

Professional journey and achievements

Stealth Startup (Restaurant Tech)

Founding Machine Learning Engineer

Seattle, USA
Nov. 2025 – present
  • Led end-to-end system design for a restaurant tech product, translating CEO-level requirements into full technical architecture spanning ML pipelines, Django backend services, and database design.
  • Coordinated and directed a small engineering team, delegating tasks, defining implementation standards, and making key technical decisions across the full stack.
  • Designed the complete database schema and table relations to support all product functionality from scratch.

GetMySpace (Parking Management Startup)

Machine Learning Engineer

Ahmedabad, India
Jan 2025 – Jun 2025
  • Developed GetMySpace's first AI-powered prototype for real-time parking management, integrating computer vision and machine learning to automate vehicle and slot detection
  • Built and deployed models using Python and PyTorch, reducing end-to-end processing time to ~4 seconds and achieving real-time update latency of ~0.7 seconds
  • Designed and implemented backend logic and MongoDB-based systems to handle prediction outputs, user logs, and dynamic updates, ensuring scalable and efficient performance

My Research

Research work in machine learning and data-driven solutions

A Two-Stage, Leakage-Aware Framework for Early Academic Risk Detection in Undergraduate Engineering Cohorts

Sep 2025
  • Proposed and implemented a leakage-aware, two-stage ML framework to forecast student academic decline
  • Applied leakage-aware modeling to ensure predictions are temporally valid and unbiased
  • Evaluated 50+ regression pipelines per subject (≈200 total) to benchmark academic performance forecasting methods
  • Tested 50+ classification pipelines for risk detection; best-performing stacking ensemble achieved recall 0.657
  • Applied explainability (SHAP) to support model interpretability and trust
Student Performance Prediction Early Warning Systems Ensemble Learning Stacked Classification Leakage Prevention Model Explainability (SHAP)
Read Paper →

My Projects

Innovative solutions powered by Machine Learning

Time-Turner: Alarm Clock for the Visually Impaired

Mar 2026
  • Designed an accessible alarm clock for blind users using physical dials as input, tactile markers, no app or wifi required
  • Defined persona, user flow, design decisions, risk analysis, and go-to-market roadmap across an 11-page case study
  • Core insight: clock hands as input mechanism rather than output, eliminating the need for a touchscreen or voice entirely
  • Dual-audience design works for both blind and sighted users on the same device with no compromises
Product Thinking Accessibility Hardware UX User Research Go-to-Market
View Case Study →

PredictGrad

Jul 2025
  • Built a full ML pipeline to predict Semester 3 marks and detect academic risk
  • Regression: Voting Regressor (Ridge + Lasso + ElasticNet)
  • Classification: Stacking (CatBoost, LGBM, ExtraTrees)
  • Deployed as interactive Streamlit app with XAI explanations
Python Scikit-learn SHAP Streamlit
Try It Live →

Falsifi.AI

Aug 2025
  • Verifies scanned signatures as genuine or forged using deep learning-based handwriting analysis.
  • Runs ~5 experimental pipelines combining distinct preprocessing and model architectures for robustness.
  • Signer-disjoint train/val/test on the CEDAR dataset to evaluate on unseen identities.
  • Streamlit UI for uploading an anchor and a candidate; outputs label, confidence, and GradCAM/SHAP explanations.
Python PyTorch CNNs ResNet TorchVision
Try It Live →

Constitution Preamble Clustering

Oct 2025
  • Utilized NLP and clustering (SentenceTransformer, KMeans) to analyze global constitutional texts and identify data-driven thematic clusters.
  • Applied semantic embeddings and unsupervised clustering to uncover ideological similarities across constitutions.
  • Built an interactive Streamlit dashboard with PCA and Plotly for visualization, interpretability, and exploration of global constitutional themes.
  • Mapped nations into clusters such as Legal-Institutional, Unity & Pluralism, and Faith & Founding, offering insight into shared governance philosophies.
Python Sentence-Transformers scikit-learn PCA Plotly Streamlit
Try It Live →

Beyond The Marks

Mar 2025
  • ML pipeline with explainable AI to analyze student performance
  • Detects grading bias and assesses influence of sensitive attributes
  • Quantified teacher effectiveness with performance indicators
  • Statistical analysis of attendance, grades, and student features
Python SHAP Pandas Matplotlib
Try It Live →

Certificates

My certifications and achievements

Skills & Expertise

Technologies I work with

Education

Academic journey and learning path

Bachelors of Engineering in AI & ML

LJ University, India

Semester 1 - Subjects
Java Software Engineering Mathematics-1 Physics IoT Computer Hardware
Semester 2 - Subjects
Advanced Java PostgreSQL MySQL Mathematics-2 Data Structures Electronics
Semester 3 - Subjects
Web Development Python Probability & Statistics
Semester 4 - Subjects
Theory of Computation Computer Organization Discrete Mathematics MERN Stack Supervised ML Advanced Python
Semester 5 - Subjects
.NET Aptitude Economics & Finance Computer Network DAA Technical English

Let's Work Together


I'd love to hear about your project and how I can help with ML solutions.

I'm always excited to work on innovative projects that make a real impact. Whether you need ML solutions, data analysis, or want to explore AI possibilities, let's connect and create something amazing together.

Send Email LinkedIn GitHub