Hi 👋, I'm Venkata Sai Teja Mothukuri

AI/ML Engineer | 3+ Years Experience | Deep Learning • Generative AI • MLOps

---

🚀 About Me

• 🤖 AI/ML Engineer with 3+ years building and deploying ML, deep-learning, and Generative-AI systems end to end
• 🧠 I train multi-modal deep-learning models (PyTorch, Hugging Face Transformers) and ship them to real-time edge & cloud inference
• 🔎 Build RAG / LLM pipelines — embeddings, vector search, semantic retrieval, transformer NLP — for production
• ⚙️ Own the full lifecycle: data & experimentation → training → deployment → MLOps, monitoring & drift detection
• 🎥 Creator of TechRex – content on AI, ML & Automation

---

💼 What I Do

• 🧩 Build multi-modal DL models (vision + text + speech) and optimize them for low-latency edge inference
• 🔮 Ship RAG & LLM features — embeddings, vector databases, semantic search, prompt engineering
• 📈 Design recommendation & ranking systems at scale (XGBoost, LightGBM, LSTM, PySpark)
• 🚀 Stand up MLOps pipelines — CI/CD for models, drift detection, model versioning (AWS, Kubernetes, MLflow)
• 🧪 Run rigorous A/B experiments with causal inference

---

🛠️ Tech Stack

👨‍💻 Languages & Frameworks

🧠 ML / Deep Learning / GenAI

• PyTorch, TensorFlow / Keras, scikit-learn, XGBoost, LightGBM
• Transformers, Bi-LSTM, CNNs, Transfer Learning, Recommendation Systems
• GenAI: RAG, LLMs, Embeddings, Vector Databases, Semantic Search, LangChain, Prompt Engineering
• NLP / CV / Speech: Hugging Face, OpenCV, NLTK, Speech-to-Text
• Data & MLOps: PySpark, Kafka, MLflow, FastAPI, Docker, Kubernetes, AWS (S3 / EC2 / Lambda / SageMaker)

---

🚀 Featured Projects

📌 Multi-Modal Cyberbullying Detection — PyTorch, Transformers, OpenCV

• End-to-end multi-modal DL system classifying text + image + audio into 6 harm categories
• Bi-LSTM text classifier 83.4% acc (0.83 F1) · VGG16 image model 96.1% acc (0.96 F1)
• Speech-to-text audio path; reproducible training (stratified splits, early stopping, confusion-matrix eval)
• 🔗 https://github.com/saiteja007-mv/Cyberbully-Detection-Multimodal-PyTorch

---

📌 Intelligent Real-Time Traffic Flow Prediction — LSTM, MLOps

• Real-time traffic prediction with LSTM: +15–20% accuracy, <2s inference for smart-city signal optimization
• End-to-end MLOps pipeline: data preprocessing → model training → interactive dashboards

---

🌱 Currently Exploring

• 🧠 LLM fine-tuning, agentic AI & retrieval-augmented systems
• ☁️ Scalable model serving & inference optimization (AWS SageMaker, on-device / edge)
• ⚙️ MLOps tooling (MLflow, drift detection, CI/CD for ML)

---

🌐 Connect With Me

---

📫 Contact

• 📧 [email protected]
• 🌐 Portfolio: https://saitejamothukuri.com

---

📊 GitHub Stats