Pawan S.
CAPTCHA Recognition using CRNN
Experience
Apr 2025 - Aug 2025
5 monthsCAPTCHA Recognition using CRNN
- Built a CRNN model with VGG16 and BiLSTM backbone for text-based CAPTCHA recognition
- Achieved 9.37% character error rate and 68.36% sequence accuracy on validation data
- Expanded data augmentation pipeline with distortions, noise injection, and clutter to improve robustness
- Conducted detailed error analysis on confusable characters (O/Q/D) and proposed error-specific augmentation
- Tech Stack: Python, TensorFlow/Keras, OpenCV, NumPy, Matplotlib
Apr 2025 - Aug 2025
5 monthsLoRA Style Fine-Tuning with Stable Diffusion
- Implemented LoRA fine-tuning on Stable Diffusion 1.5 to generate Studio Ghibli-inspired imagery
- Optimized hyperparameters (learning rate, epochs, LoRA rank/alpha) for stable convergence and effective style transfer
- Designed an evaluation framework with negative prompting and systematic seed testing to ensure consistency
- Managed training checkpoints and prompt engineering for reproducible high-quality outputs
- Tech Stack: Python, PyTorch, Hugging Face Diffusers, Stable Diffusion 1.5
Apr 2025 - Aug 2025
5 monthsTranslation Bias Analysis with LLMs
- Investigated age and gender bias in German–English translation across MT tools and LLMs (GPT-4, DeepSeek)
- Implemented Logistic Regression, SVM, and multilingual BERT classifiers for demographic prediction
- Designed and tested baseline, preserved, and adaptive prompting strategies for bias mitigation
- Contributed training and inference code for LR and SVM, visualization of gender and age bias, and report methodology
- Tech Stack: Python, scikit-learn, Hugging Face Transformers, Logistic Regression, SVM, BERT
Oct 2024 - Feb 2025
5 monthsStock Market Prediction System
University of Technology Nuremberg
- Engineered an ML pipeline integrating Yahoo Finance data with real-time news sentiment analysis
- Implemented feature engineering combining technical indicators with FinBERT sentiment scores
- Achieved 0.94 ROC AUC using an optimized ensemble of Naive Bayes, Logistic Regression, and SVM
- Tech Stack: Python, scikit-learn, pandas, numpy, matplotlib, seaborn
Sep 2021 - Nov 2024
3 years 3 monthsHyderabad, India
Senior Data Analyst
Tiger Analytics
- Designed and implemented customer segmentation models for retail clients, reducing the target customer base by 70% while maintaining 95% revenue coverage
- Migrated legacy SAS workflows to MySQL, improving processing speed by 40% and reducing operational costs for an insurance client
- Developed insurance sales prediction models using H2O AutoML, achieving over 90% accuracy
- Built scalable PySpark pipelines on Databricks for processing 50M+ records, reducing processing time from 8 hours to 45 minutes
- Delivered comprehensive patient journey analysis for a Fortune 500 healthcare client, creating automated Power BI dashboards tracking 15+ KPIs
- Implemented real-time data pipelines using Snowflake and Dataiku
- Created interactive dashboards serving over 200 stakeholders
- Developed drug-switching analysis tools for pharmaceutical research
Jan 2020 - Sep 2021
1 year 9 monthsNoida, India
Technical Content Engineer
GeeksforGeeks
- Authored and reviewed over 110 technical articles on state-of-the-art artificial intelligence topics
- Developed production-ready code examples for complex algorithms such as BERT, GPT, YOLO, and ResNet
- Created educational content on computer vision, NLP, and deep learning fundamentals
Aug 2019 - Dec 2019
5 monthsHuman Activity Recognition for Patient Monitoring
- Developed a real-time action detection system for hospital patient safety monitoring
- Implemented OpenPose for pose estimation and DeepSORT for multi-object tracking
- Achieved 95% accuracy in fall detection with less than 200 ms latency
- Tech Stack: Python, OpenCV, TensorFlow, Keras
Jan 2019 - Jun 2019
6 monthsIntelligent Attendance System with Face Recognition
- Built a frame-wise face recognition system, achieving around 97% accuracy
- Extended the solution for UAV surveillance applications
- Published “UAV Surveillance for Violence Detection and Individual Identification”
- Tech Stack: Python, OpenCV, TensorFlow, Keras, computer vision libraries
Summary
Results-driven Artificial Intelligence and Machine Learning Engineer with 3+ years of professional experience in machine learning, deep learning, and advanced analytics.
Proven track record of delivering scalable ML solutions for Fortune 500 clients, with expertise in end-to-end data science pipelines, customer segmentation, and predictive modeling.
Languages
English
AdvancedGerman
ElementaryEducation
Oct 2024 - Oct 2026
University of Technology (UTN)
Master of Science · Artificial Intelligence & Robotics · Nuremberg, Germany
Aug 2016 - Jun 2020
Bennett University
Bachelor of Technology · Computer Science & Engineering · Greater Noida, India · 9.28/10 (Equivalent to German Grade: 1.3 - Very Good)
Certifications & licenses
Large Language Models: Foundation Models from the Ground Up
Large Language Models: Application through Production - Databricks
AWS Certified Cloud Practitioner
Google Data Analytics Specialization
Need a freelancer? Find your match in seconds.
Try FRATCH GPT More actions
Similar Freelancers
Discover other experts with similar qualifications and experience
- English
- Deutsch
2025 © FRATCH.IO GmbH. All rights reserved.