Hi, I'm Bhavya

Dedicated AI enthusiast and penultimate-year student at Madhav Institute of Technology and Science, Gwalior, pursuing a Bachelor of Technology in Artificial Intelligence and Machine Learning. I have a strong foundation in Gen AI solutions, data analysis, transformer-based model development, and production-level backend development. My interests extend to open-source contributions, competitive programming in C++ (CodeChef 3-star), and participating in hackathons and developer communities.

Experience:

Artificial Intelligence Intern | FoodNEST(S)

• Goal: Enhance food delivery supply chain efficiency through AI solutions.
• Key Contributions:
  • Devised and deployed a custom in-house multilingual, multimodal AI chatbot, automating order processing for 2M+ stakeholders and reducing manual management by 40%.
  • Developed a proprietary multilingual architecture, outperforming existing models by effectively supporting 22+ Indic languages.
  • Engineered and deployed web, map, and video scrapers hosted on EC2 with Gunicorn and Docker for scalable data collection from leading e-commerce platforms.

Artificial Intelligence Engineer | Radical X

• Goal: Advance AI-driven career solutions and mentor systems.
• Key Contributions:
  • Enhanced AI solutions like Rex (AI coach) using OpenAI, Vertex AI, and TensorFlow.
  • Collaborated on multiple hackathons and developed personalized career guidance systems.

Software Engineering Fellow | Headstarter AI

• Goal: Build and promote enterprise AI projects.
• Key Contributions:
  • Created a multimodal voice assistant and a music AI tool with visual generation, gaining significant community traction.
  • Promoted projects, leading to client engagement and brand visibility.

Event Coordinator | Developers

• Key Contributions:
  • Coordinated online webinars, boosting community engagement by 78% through targeted email marketing.
  • Established efficient pipelines for seamless event execution.

Projects:

1. Abstractive Gen AI Q&A Bot:

• Goal: Develop an enhanced question-answering system with optimized vector storage and efficient all-dataset retrieval capabilities.
• Solution: Built a Q&A system with the ELI5 BART model, integrated Pinecone for vector storage, and used Flax embeddings for retrieval, reducing query latency from 500ms to 50ms.

2. YouStock:

• Goal: Develop trading strategies and provide personalized investment insights.
• Solution: Analyzed market dynamics and quarterly earnings with Prophet and integrated sentiment analysis to improve precision by 10%. Deployed insights via Streamlit, boosting latency by 27%.

3. Project Narmadae:

• Goal: Detect water usage anomalies and forecast demand effectively.
• Solution: Achieved 84% F1 score for leak detection using Isolation Factor, One-Class SVM, and Local Outlier Factor. Forecasted water demand with 89% accuracy using PyCaret.

4. Neural Machine Translation with Fairseq:

• Goal: Develop a machine translation model on custom data.
• Solution: Utilized Fairseq NMT model with IIT Bombay English-Hindi Dataset. Evaluated using SACRE BLEU and BLEU metrics.

5. RAG-Enhanced Chatbot:

• Goal: Develop a highly responsive chatbot using RAG, LangChain, and ChromaDB for dynamic conversational interactions.
• Solution: Integrated RAG, LangChain, ChromaDB, and GPT-3.5 Turbo for enhanced responsiveness.

Technical Skills:

• Languages: Python, C++, JavaScript, SQL
• Developer Tools: TensorFlow, PyTorch, Hugging Face Transformers, LangChain, Flask, FastAPI, Docker, AWS, GCP
• Techniques and Concepts: Data Analysis, Time Series Analysis, Deep Learning (CNN, RNN, LSTM, GANs), NLP, NMT, LLM, MLM, Text Generation, Sentiment Analysis, LLaMA & OpenAI Fine-tuning
• Libraries: pandas, NumPy, Matplotlib, Seaborn, Scikit-learn

Certifications:

• Coursera: Machine Learning Specialization by Andrew Ng, Stanford
• Coursera: Natural Language Processing Specialization by DeepLearning.AI
• Finalist at Wells Fargo Hackathon Conducted by IIT Madras