Skip to content
View SheemaMasood381's full-sized avatar
🏠
Working from home
🏠
Working from home

Block or report SheemaMasood381

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 supported. This note will be visible to only you.
Report abuse

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

Report abuse
SheemaMasood381/README.md

Business Card

Building with purpose. Learning with obsession. Solving with AI.

🚀 Automating the Future!!!! ' From LLM Pipelines to Autonomous Workflows '

I'm an AI Automation Engineer at the intersection of machine learning, agentic systems, and workflow intelligence — passionate about transforming ideas into autonomous, scalable solutions.

Whether it’s orchestrating LLM-driven workflows, building n8n automations, or crafting intelligent agents, I thrive on creating systems that think, act, and evolve.


Python TensorFlow PyTorch RAG Vertex AI MistralAI OpenAI Streamlit CrewAI LangChain Solana Docker GitHub Actions Lablab.ai Kaggle Hackathons

🔍 About Me

💡 AI Automation Engineer passionate about LLMs, agentic systems, and intelligent automation.
I specialize in connecting data pipelines, APIs, and AI models using n8n, Python, and LangChain — crafting automations that enhance productivity and decision-making.

🧠 My expertise lies in:

  • Building agentic multi-agent ecosystems using CrewAI, LangChain, and LLMs.
  • Designing end-to-end automation workflows (document processing, chatbots, AI assistants).
  • Engineering Generative AI pipelines for text, voice, and data-driven tasks.
  • Deploying scalable AI systems with Docker, Vertex AI, and GitHub Actions.

With a Pharm.D background, I bring analytical precision and real-world empathy into AI — ensuring every system I build matters.


⚙️ AI Automation Workflows

Integrating n8n + LLMs + Python for business process automation

  • Developed document parsing pipelines, voice-to-doc automation, and report generators.
  • Created autonomous task orchestration systems using multi-agent logic.
  • Reduced manual workflow time by 90% through intelligent automation.

💭 What Drives Me

I’m fueled by curiosity and precision — continuously learning, building, and deploying AI that transforms real-world workflows.

🤝 My mission:

  • To build autonomous systems that scale effortlessly.
  • To collaborate on AI-first startups creating meaningful automation.
  • To enable humans + machines to work together intelligently.

🏆 Featured Achievements

  • 🚀 CheapestBuy.AI — Built at Internet of Agents Hackathon @Solana Skyline (lablab.ai, Solana, Coral MCP, Mistral AI, AIMLAPI)

    • Registered shopping agent on Coral MCP
    • Integrated CrewAI multi-agent workflows, RAG, Solana Pay (DEVnet demo) + Helius API for on-chain verification
    • Added voice search (Whisper GPT-4o) for multilingual queries
    • 🎖️ Project Spotlight by AIMLAPI
    • 🎤 Invited to present at AIMLAPI Shark Tank (Discord livestream)
  • 🏅 Hackathon Finalist & Multiple Top Ranks on Kaggle

    • Petals to the Metal (3rd place, official Kaggle competition)
    • Blood Glucose Prediction (Top 15%)
    • Detect AI vs Human Images (Top 25 public LB, invited to share methodology)

🚀 Featured Projects

🧠 Multi-agent shopping assistant using CrewAI, Gemini, Whisper, Serper API, and LiteLLM.

  • Personalized recommendations and product summarization via RAG
  • Real-time product parsing & comparison with voice and text support
  • Built in Streamlit with 4-star expert rating at Lablab’s Execute Hackathon

🥇 Ranked 25th on public LB, 29th on private LB with F1 score: 0.80

  • Used EfficientNet, ViT, and custom augmentations
  • Selected among standout solutions; requested to present methodology due to high-quality results

🥉 Secured 3rd position in the official Kaggle "Petals to the Metal" competition using macro-F1 evaluation.

  • Built a high-performing ensemble using ConvNeXtBase, InceptionResNetV2, and ResNet50
  • Performed grid search over weights (α, β, γ) to maximize F1 score
  • Achieved F1 score = 0.9568 on first submission 🚀
  • Visualized results with confusion matrix, classification report, and model comparison charts

🎯 Achieved Top 19% (Rank 378 / 2025) with 87% accuracy

  • Custom mel-spectrogram pipeline using seresnext26t_32x4d
  • Advanced TTA, chunking, and ensemble inference
  • Addresses bioacoustic species detection in real-world environments

📊 Ranked in Top 15% with RMSE = 0.09

  • Time-series prediction using XGBoost, LGBM, and stacking
  • Engineered temporal features for BGL variation modeling

🧬 Achieved Top 30% in RSNA competition

  • Medical image classification using ResNet + EFFICIENTNET
  • Emphasis on class balancing and clinical impact

🗣️ Real-time voice chatbot using Whisper, Llama3, and gTTS

  • Reduces typing effort by 90%, ideal for customer support
  • Conversational, human-like interaction via Gradio interface

📚 Document Q&A and summarization using Gemini + ChromaDB

  • Extracts insights from PDFs, saving 70% review time
  • Gradio UI for instant document interrogation

🔎 Cuts down reading time by 80%

  • Summarizes PDFs, Word, and TXT files using Gemini + Gradio
  • Ideal for researchers and students needing instant insights

🎭 A fun NLP project fine-tuning LLMs for Yoda-style speech generation

  • Entertainment + creative writing use case
  • Deployed with Gradio for live demo interaction

🛠️ Technical Skills & Tools

👩‍💻 Languages & Frameworks

  • Python, SQL, JavaScript (Basics)
  • PyTorch, TensorFlow, Scikit-learn
  • LangChain, CrewAI, FastAPI, Flask, Streamlit

🤖 AI, LLMs & Agentic Systems

  • LLMs (OpenAI, Gemini, LLaMA, Mistral)
  • LangChain Agents, CrewAI Multi-Agent Workflows
  • RAG Pipelines, Vector DBs (ChromaDB, FAISS, Pinecone)
  • Coral Protocol (MCP), Solana Pay AI Agents
  • Whisper, Voice AI, Speech-to-Text / TTS
  • Vision Transformers (ViTs), CNNs, Transfer Learning
  • Fine-tuning, Prompt Engineering, Generative AI
  • Explainable AI (SHAP, LIME)

📊 Data Science & Machine Learning

  • Regression, Classification, Time Series Forecasting
  • Feature Engineering, Model Optimization, Cross Validation
  • EDA, Statistical Analysis, Data Wrangling

🖼️ Visualization & BI Tools

  • Power BI, Matplotlib, Seaborn, Plotly
  • Gradio, Streamlit, Interactive Dashboards

⚙️ Tools, Libraries & Tech

  • OpenCV, Pandas, NumPy
  • Hugging Face, Serper API, LiteLLM
  • API Integration, Agent-API Orchestration

💻 DevOps, Deployment & Cloud

  • Docker, GitHub Actions (CI/CD)
  • Render, Hugging Face Spaces, Vertex AI
  • Jupyter, Colab, VS Code, Cursor

🧪 MLOps & Experimentation

  • Experiment Tracking, Model Versioning
  • Feature Stores, Model Registries
  • Kaggle Notebooks & Competitions

📜 Certifications


Let's Connect:


GitHub Stats

Sheema's GitHub stats


Top Languages

Top Languages


“Automation is not the end of work — it’s the evolution of intelligence.” Thanks for stopping by!
Let’s connect, collaborate, and build something extraordinary together. 🚀

Pinned Loading

  1. Pearls-Karachi-AQI-Prediction-for-next-3-days- Pearls-Karachi-AQI-Prediction-for-next-3-days- Public

    End-to-end AQI Prediction System using real-time weather & pollutant data, with automated ML pipelines and a Streamlit dashboard.

    Jupyter Notebook 1 1

  2. Execute_Ai_Genesis_Hackathon_Lablabai_april2025 Execute_Ai_Genesis_Hackathon_Lablabai_april2025 Public

    ShopSmart.Ai - An intelligent shopping Assistant built with CRewAi

    Jupyter Notebook 2

  3. Detect_AI_vs_Human_Generated_Image Detect_AI_vs_Human_Generated_Image Public

    A deep learning-based model to distinguish between AI-generated and human-created images. This repository includes data preprocessing, model training (ConVitX with ConvNeXt and Swin Transformer), e…

    Jupyter Notebook 1

  4. Blood-Glucose-Levels-Prediction-2024-kaggle-competition Blood-Glucose-Levels-Prediction-2024-kaggle-competition Public

    "Predicting blood glucose levels using advanced machine learning techniques, including XGBoost, LightGBM, CatBoost, Random Forest, TabNet, and model stacking/ensembling with K-Fold cross-validation…

    Jupyter Notebook 2 1

  5. BirdClef-2025-Kaggle BirdClef-2025-Kaggle Public

    BirdCLEF 2025 soundscape classification project using deep learning (SeresNeXt26t) with PyTorch, pseudo-labeling, and audio preprocessing for bird species identification.

    Jupyter Notebook 1

  6. Kaggle-Competitions Kaggle-Competitions Public

    My Kaggle competition notebooks, analysis, and ML solutions , updated regularly with new challenges.

    Jupyter Notebook 1