AI / Machine Learning
PyTorchTensorFlowScikit-learnXGBoostHuggingFace
AI/ML Engineer Portfolio
Satyam Mishra
AI/ML Engineer | Generative AI | RAG Systems | MLOps
I build end-to-end AI systems from data pipelines to production deployment with a focus on scalable machine learning and generative AI.
About
Satyam Mishra is a final-year Artificial Intelligence and Machine Learning student with strong experience in building end-to-end machine learning systems. His work spans NLP, generative AI, MLOps pipelines, and scalable cloud deployment using AWS and Docker. He has also conducted research in fake news detection and built simulation models exploring demographic and economic dynamics.
Skills
Generative AI
RAG SystemsLangChainFAISSOpenAI Whisper
MLOps
DockerGitHub ActionsMLflowDVC
Cloud
AWS EC2AWS S3AWS ECRMongoDB Atlas
Backend
FastAPIStreamlitPython
Featured Projects
Local-First YouTube RAG System
A privacy-first retrieval augmented generation system that enables contextual conversations with YouTube content using local transcription and vector search.
LangChainFAISSWhisperPython
- Local speech transcription
- Vector similarity search
- Context-aware responses
- Cost-efficient local processing
Vehicle Insurance ML System
An end-to-end MLOps pipeline for vehicle insurance prediction including automated data ingestion, model training, and cloud deployment.
FastAPIDockerAWS EC2MongoDB AtlasGitHub Actions
- Modular ML pipeline
- CI/CD automation
- Dockerized deployment
- Cloud infrastructure integration
ATS Resume Analyzer
An NLP-based resume scoring system that measures semantic similarity between resumes and job descriptions.
SBERTTF-IDFXGBoostStreamlit
- Semantic similarity analysis
- Resume-job matching
- Interactive scoring interface
Demographic Inversion Simulation
A computational simulation analyzing the economic and workforce implications of a hypothetical one-child policy in India.
PythonData AnalysisEconomic Modeling
- Workforce projection simulation
- Fiscal balance modeling
- Dependency ratio analysis
- Data-driven policy exploration
Research
Fake News Detection Using Logistic Regression - ICAMC 2024
Feb 2024 - May 2024
- Led research on fake news detection using Logistic Regression, achieving 94% training accuracy and 93% testing accuracy, resulting in a peer-reviewed publication presented at ICAMC 2024.
- Developed a complete NLP pipeline using scikit-learn, including lemmatization, stopword removal, and TF-IDF vectorization to convert unstructured text into structured features.
- Built a binary classifier to detect misinformation on social media and evaluated performance using F1-score, confusion matrix, and accuracy metrics.
- Co-authored and presented research at an international conference, contributing to AI/ML applications in misinformation detection.
94% training accuracy93% testing accuracyICAMC 2024
GitHub
Repositories
43
Stars
38