Ravula Lohith – AI/ML Enthusiast | Open-Source Contributor | Builder

🧑‍💻 About Me

I'm Ravula Lohith, a B.Tech CSE (AI & ML) student passionate about open-source AI, real-world ML deployment, and human-AI collaboration. I believe in building tools that are efficient, impactful, and grounded in practical use cases.

🔧 Internship Projects – Swecha

1. 📄 Transcription – from Regional YouTube Videos

• Repo: video_wav.ipynb

• Automated transcription pipeline for Creative Commons Telugu YouTube videos:

  • Downloaded videos using yt-dlp
  • Converted to .wav using ffmpeg
  • Transcribed using Whisper and Gemini APIs

• ✅ Automated 90% of manual transcription work, enabling domain-level NLP processing

2. 🧠 Fine-Tuning LLMs using Unsloth

• Repo: finetuning

• Fine-tuned LLaMA models with Unsloth (QLoRA-based):

  • Trained on custom instruction datasets
  • Enabled text generation use cases

• ✅ Achieved 25% faster training and 18% better prompt relevance

3. ✈️ SkyTrip – AI-Powered Travel Assistant

• Repo: SkyTrip

• AI travel planner using Gemini API, Langchain, Python, and Streamlit:

  • Displays current weather, top 5 tourist places (with Google Maps integration)
  • Recommends what to carry, what to play, etc.

• ✅ End-to-end intelligent assistant for travel planning in <10s

4. ⚙️ SOAI Prerequisites Automation

• Repo: SOAI__automation.py

• Developed an OS-aware script to automatically:

  • Install Python, Git, Streamlit, JupyterLab, VSCodium
  • Set up virtual environments and validate dependencies

• ✅ Reduced setup time from 30 minutes to <5 minutes, streamlining contributor onboarding

5. 🤖 Personal AI Assistant – Prompt Engineering with Hugging Face

• Repo: Hugging Face Conversation

• Built an Interview Preparation Assistant using open-source LLMs:

  • Accepts user skills as input and generates custom Q&A sets
  • Explored prompt engineering techniques including few-shot, zero-shot, and persona shaping

• ✅ Designed for future integration into productivity and upskilling tools

As a volunteer at AIDAYS 2025,

• Assisted in organizing and coordinating AI-centric workshops
• Gained insights into deploying AI solutions in grassroots scenarios
• Collaborated with experts and peers on open-source-driven innovation

💡 Personal Projects

🏠 House Price Prediction

• Repo: GitHub
• Used Linear Regression to predict house prices based on features like size, location, etc.
• ✅ Performed EDA, feature engineering, and model evaluation
• 💡 Useful for real estate pricing insights and financial planning tools

📱 Mobile Phone Activity Classification

• Repo: GitHub
• Classified phone usage behavior (e.g., homework, personal, or entertainment) using browsing activity data
• Applied time-series feature extraction and classification techniques
• ✅ Can be used in parental control systems and usage monitoring tools

🚀 Tools & Technologies

• Languages: Python, Bash,SQL
• Frameworks/APIs: Whisper, Hugging Face Transformers, Gemini API, Langchain, Streamlit, ffmpeg
• Libraries: NumPy, Pandas, scikit-learn
• Concepts: Prompt Engineering, Text Generation, LLM Fine-Tuning, Regression, Classification
• Collaboration: Git, GitLab

📬 Contact

• 📧 Email: [email protected]
• 🔗 LinkedIn: Ravula Lohith
• 🐙 GitHub: RavulaLohith