Skip to content

Jahnavi Pendyala

Machine Learning Engineer • Ankampalem, Andhra Pradesh, India • p*************@gmail.com • 911****999 • linkedin.com/••••• • drivetube.ai/•••••

Career Objective

Machine Learning Engineer with 0 years of experience designing and delivering NLP and computer-vision solutions. Built BERT-based resume screening pipelines and YOLOv5-based defect detection models; proficient in Python, SQL, PyTorch, Hugging Face Transformers, OpenCV, and scikit-learn. Skilled at turning data into production-ready models, reducing manual effort, and improving matching and detection accuracy. Seeking an entry-level ML/AI or data science role where technical and analytical skills can drive measurable business impact.

Education

Sasi Institute
Bachelor of Technology (B.Tech) – Computer Science and Engineering • Andhra Pradesh, India • 2021 – 2025

Technical Skills

Programming Language: Python,Java,JavaScript,SQL,Data Structures,Software Engineering,Operating Systems,Networking TCP,Object-Oriented Programming
Frontend Technologies: HTML,CSS
Databases: MySQL,DBMS
Version Control & Development Tools: Git,Visual Studio Code,Windows
Data Analysis & Visualization: Pandas,NumPy,Microsoft Excel,PowerPoint
Machine Learning & AI: Machine Learning,Computer Vision,Natural Language Processing
AI/ML Frameworks & Libraries: PyTorch,Hugging Face Transformers,scikit-learn,OpenCV
Generative AI & LLMs: ChatGPT,GitHub Copilot
Networking & Infrastructure: DNS,DHCP
Project Management & Collaboration: Microsoft Word,Agile,YOLOv5

Projects

Intelligent Resume Screening Using NLP & BERT
Tools Used: Python, Hugging Face Transformers, scikit-learn, pandas, SQL
  • Engineered a BERT-based semantic matching pipeline to process 500+ resumes, implementing cosine-similarity ranking that achieved 92% top-match accuracy.
  • Built end-to-end NLP preprocessing (tokenization, entity extraction, skill parsing) using Python and Hugging Face to parse 15+ skill categories and normalize resume fields.
  • Automated candidate ranking and report generation with pandas and SQL, reducing manual screening time by ~4 hours per recruitment cycle (≈60% reduction).
  • Designed evaluation metrics and validation workflow; improved match precision by 30% over keyword-based baselines and achieved an 89% F1-score across test sets.
  • Integrated scikit-learn classifiers for post-filtering and threshold tuning to balance recall and precision for different role profiles.
Solar Cell Defect Detection using YOLOv5
Tools Used: Python, YOLOv5, PyTorch, OpenCV, NumPy
  • Trained and fine-tuned a YOLOv5 object detection model on a dataset of 1,000+ images to identify cracks, hotspots, and broken cells, achieving 92% detection accuracy on the test set.
  • Applied data augmentation and transfer learning strategies to increase robustness, improving detection reliability by 25% on augmented validation splits.
  • Implemented an optimized real-time inference pipeline using PyTorch and OpenCV, delivering 15 FPS video processing for on-line inspection scenarios.
  • Reduced false positives by 20% through threshold tuning, non-maximum suppression adjustments, and post-processing filters tailored to industrial imagery.
  • Created model evaluation dashboards with precision/recall curves and confusion matrices to guide iterative improvements and deployment readiness.

Certifications

Cloud Foundations — AWS Academy
Cyber Security & Cyber Essentials — Cisco Networking Academy

Powered by Drivetube · Create your own profile at drivetube.ai