- Led a team of 4 in completing an LLM project for a Japanese client.
- Built Graph DB, deployed FastAPI, and installed cloud driver.
- Developed and deployed LLM models like Gemma, Mamba, Vinal-lamma, and GPT.
- Utilized Neo4j to store and manage graph data, ensuring efficient query performance.
- Implemented Retrieval-Augmented Generation (RAG) to extract information from forms.
- Integrated Pinecone for vector database storage, enabling fast and scalable similarity searches.
- Developed advanced Speech Recognition systems using deep learning models such as LSTMs and Transformer architectures.
- Built a Speech Emotion Recognition system leveraging audio features like Mel-frequency cepstral coefficients (MFCC) and spectrograms to classify emotions such as happiness, frustration, and neutrality in speech data.
- Applied Transformer-based models such as Wav2Vec2.0 for speech processing, enhancing real-time understanding of spoken language.