Skip to content
View adityasinghcoding's full-sized avatar
:octocat:
Let's collaborate on projects.
:octocat:
Let's collaborate on projects.

Block or report adityasinghcoding

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
adityasinghcoding/README.md
Typing SVG

LinkedIn Portfolio Email ORCID


🧠 "I don't just build AI systems β€” I ask why they behave the way they do, then optimise them."


πŸ‘¨β€πŸ’» About Me

I'm an AI/ML Engineer from Moradabad, UP, India β€” obsessed with turning research-grade ideas into production-grade systems. My work spans the complete ML lifecycle: raw data ingestion, model training, prompt engineering, RAG pipeline design, agentic orchestration, and end-to-end deployment.

  • πŸš€ Shipped: PolySensor β€” a live agentic AI platform processing 108+ file formats with RAG-powered content intelligence
  • πŸ”¬ Shipped: ACMA β€” a deployed multimodal content moderation system (F1: 0.91)
  • 🧩 Advisory: Designed enterprise AI architecture across 4 business functions for CL Gupta Exports Ltd.
  • πŸ“– Research habit: I read ML papers and translate findings directly into project implementations
  • πŸ₯‡ Off the screen: 8Γ— Gold Medalist in Speed Skating β€” discipline on ice, precision in code

πŸ› οΈ Tech Stack

aditya = {
    "LLMs & Agentic AI"   : ["LangChain", "LangGraph", "Google Gemini API", "OpenAI API", "Hugging Face"],
    "ML & Deep Learning"  : ["PyTorch", "TensorFlow", "Keras", "Scikit-learn", "CNN", "YOLOv8"],
    "NLP & Multimodal"    : ["RAG", "ChromaDB", "OCR (Tesseract/EasyOCR)", "Speech Recognition", "NER"],
    "Data & Pipelines"    : ["Pandas", "NumPy", "Matplotlib", "Seaborn", "ETL Design", "SQL/SQLite"],
    "Backend & Deployment": ["Flask", "FastAPI", "Docker", "Vercel", "Render", "Nginx", "Git"],
    "Currently exploring" : ["Advanced Agentic Architectures", "LLM Fine-tuning at Scale"]
}

πŸš€ Featured Projects

πŸ”΅ PolySensor Β· Live β†—

Agentic AI Β· Multimodal Β· Production Deployed

End-to-end agentic pipeline that ingests and analyses 108+ file formats β€” documents, images, audio, video β€” with multi-agent orchestration via LangChain.

  • 🧩 Advanced RAG with ChromaDB vector store + custom chunking
  • πŸŽ™οΈ OCR (Tesseract) + Speech Recognition for unstructured media
  • ⚑ 80% reduction in manual content review time
  • 🐳 Dockerised + deployed on Vercel (frontend) & Render (backend)
  • πŸ“¦ 161 commits Β· 2 releases

Python LangChain Gemini API RAG ChromaDB React Flask Docker

πŸ”΄ ACMA Β· Live β†—

AI Content Moderation Β· CNN + NLP Β· Final Year Project

Multimodal content moderation system detecting toxic, inappropriate and violent content across text, images, audio and video.

  • 🎯 F1-Score: 0.91 via systematic hyperparameter tuning
  • 🧠 TF-IDF + toxicity classifier for text; CNN (299Γ—299) for images
  • πŸŽ₯ Frame-level video analysis + speech-to-text audio pipeline
  • πŸ”Œ REST API: POST /detect_toxicity for live integrations
  • πŸ“¦ 70 commits Β· Dockerised

Python TensorFlow Keras CNN OpenCV EasyOCR Flask Docker

🟑 Enterprise AI Architecture

Advisory Β· CL Gupta Exports Ltd. Β· 2026

Comprehensive AI integration across 4 business functions for a mid-size export manufacturing firm.

  • πŸ‘₯ HR anomaly detection β€” Isolation Forest
  • πŸ“¦ Customs doc intelligence β€” LangChain + pdfplumber β†’ MS SQL Server (no LLM dependency)
  • πŸ“ˆ Demand forecasting β€” Facebook Prophet
  • πŸ‘οΈ Visual QC β€” YOLOv8
  • πŸ” Attrition prediction β€” XGBoost

XGBoost Prophet YOLOv8 LangChain pdfplumber pyodbc MS SQL Server

🟒 ML Internship @ CodSoft · Certificate

Production ML Β· 3 Domains Β· 50,000+ Records

Built and evaluated three production-ready ML models across distinct problem types.

  • πŸ“‰ Customer Churn Prediction β€” 86% accuracy
  • 🎬 Movie Genre Classification β€” 77% precision (multi-label NLP)
  • πŸ“§ SMS Spam Detection β€” 97% accuracy (binary classification)
  • ⚑ 20% reduction in training pipeline time via optimised data loading

Scikit-learn Pandas NLP TF-IDF Jupyter Β· Aug 2023


πŸ“Š GitHub Stats


πŸ… Certifications

Certification Issuer Year
🐍 Python for Data Science IBM 2023
πŸ€– Data Science & Machine Learning ShapeMySkills 2023
πŸ“ Machine Learning NPTEL / IIT Kharagpur 2022
πŸ—„οΈ Database Management Systems NPTEL / IIT Kharagpur 2023
β˜• Java Programming NPTEL / IIT Kharagpur 2021

🐍 Contribution Snake

Snake animation


🀝 Let's Connect

LinkedIn Portfolio PolySensor Email ORCID


skating on ice taught me precision Β· AI taught me scale Β· shipping taught me everything else

Visitor Count

Pinned Loading

  1. PolySensor PolySensor Public

    PolySensor is an AI-powered agentic multi-modal content analysis tool that analyzes textual information from different 108 file formats or more. Using Google Gemini and advanced RAG techniques, it …

    Python 1

  2. AI-Content-Moderation-Analysis-Final-Year-Project- AI-Content-Moderation-Analysis-Final-Year-Project- Public

    AI-driven multimodal content moderation system named ACMA to detect and analyze toxicity, inappropriate visuals, and violence in text, images, audio and video.

    HTML 1

  3. AI-Training-Assistant AI-Training-Assistant Public

    JavaScript 1

  4. CERTIFICATES CERTIFICATES Public

    Certificates of Achievements.

    1

  5. Internships Internships Public

    1

  6. Stock-Market-Operator-Game Stock-Market-Operator-Game Public

    JavaScript 1