Rushikesh Patil
ML EngineerAI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
ML Engineer with less than a year in Deep Learning & Computer Vision
AI/ML Engineer and Computer Engineering graduate skilled in building and deploying production-ready AI systems across Deep Learning, Computer Vision, and NLP. Experienced in end-to-end ML workflows using Python, PyTorch, Scikit-learn, Flask, and LangChain, spanning data preprocessing, model training, evaluation, and REST API deployment. Comfortable with transfer learning, metric learning, and CPU inference optimization for real-time applications. Passionate about AI orchestration, intelligent automation, and solving impactful real-world problems through efficient, scalable solutions.
Kai Sau G.F. Patil Kanishtha Mahavidyalaya
Higher Secondary Certificate (HSC)
N/A – May 31, 2022
Sheth V K Shah Vidya Mandir
Secondary School Certificate (SSC)
N/A – May 31, 2020
D.N. Patel College of Engineering
Bachelor of Technology · Computer Engineering
N/A – June 30, 2026
Face Kinship Verification
June 24, 2026 – Present
Built a Siamese network using AdaFace IR-101 embeddings to verify facial kinship across four relationship classes (Father-Son, Father-Daughter, Mother-Son, Mother-Daughter), trained on the KinFaceW-II and FIW datasets. Achieved 82.65% accuracy, 96.7% recall, and 84.96% F1-score under 5-fold cross-validation, keeping variance low (±4.79%) via batch normalization and dropout regularization. Designed a custom similarity head fusing Absolute Difference, Hadamard product, and Cross-Attention for robust feature comparison, with Youden's J statistic for optimal threshold calibration. Integrated MediaPipe face detection as a preprocessing gate and deployed the model via a Flask REST API with a responsive React + Tailwind frontend for real-time inference. Optimized the pipeline for CPU-only inference, enabling live kinship analysis on standard hardware without GPU dependency.
View ProjectSkin Cancer Detection
June 24, 2026 – Present
Trained a ResNet50 transfer-learning model on the HAM10000 dataset (10,000+ dermoscopic images, 7 lesion categories), reaching 83% test accuracy. Addressed severe class imbalance using Focal Loss and stratified data splitting, meaningfully improving minority-class prediction. Built a full PyTorch training pipeline with data augmentation (rotation, affine, color jitter, flips), learning-rate scheduling, and early stopping over 50 epochs. Deployed the model via a Flask REST API (/predict, /upload, /results) for real-time CPU inference and delivered a full-stack web app with drag-and-drop upload and automated risk assessment (Low / Medium / High / Critical). Evaluated performance using precision, recall, F1-score, and confusion matrices to validate reliability across all seven lesion classes.
View ProjectCultural Fit Analysis
The candidate's projects demonstrate a strong interest in applying AI/ML to real-world problems (kinship verification, skin cancer detection), which aligns well with an innovative and impact-driven culture. The use of diverse technologies (PyTorch, Flask, React, Langchain) and the open-sourcing of projects suggest a proactive and learning-oriented individual. However, the lack of professional experience or team-based projects limits the assessment of cultural fit in a collaborative work environment.
Soft Skills & Operational Fit
The candidate's project descriptions indicate a problem-solving mindset, particularly in addressing challenges like class imbalance and CPU inference optimization. The ability to build end-to-end solutions suggests a good understanding of operational workflows. However, without direct work experience or psychometric test results, it's difficult to assess stress handling, team collaboration, or communication clarity in a professional setting.