Undergraduate student in Computer Science, with a focus on Artificial Intelligence and Data Science.
Interested in building practical systems involving large language models, retrieval, and data pipelines.
-
Designed and deployed a real-time voice-based support agent with low-latency streaming (16 kHz PCM), handling order lookup, refunds, and complaint workflows via LLM tool calling.
Implemented WebSocket-based bidirectional audio streaming, live transcript updates, and backend policy logic (refund windows, escalation rules) with production deployment on GCP. -
Built a coding agent system supporting file operations, command execution, and repository-level search for multi-step software tasks.
Implemented persistent memory (todo.md,scratchpad.md,transcript.md), sliding-window context management, and grep-based retrieval to reduce token usage and improve scalability. -
Engineered time-aware data pipelines for sequential and multimodal data, enabling temporal indexing, event ordering, and trend analysis in user-generated content.
Modified backend schema and ingestion workflows to support time-dependent querying and downstream modeling. -
Developed a transformer-based sentiment analysis service (DistilBERT) for tracking narrative evolution over time.
Designed inference pipelines for continuous sentiment monitoring and integrated outputs into higher-level narrative analysis systems.
- Large Language Models (LLMs)
- Retrieval-Augmented Generation (RAG)
- Agent-based systems
- Efficient context and memory handling
- Data engineering for AI systems
Languages: Python, C++
Frameworks/Tools: Transformers, LangChain, LlamaIndex
Systems: FastAPI, Docker, GCP, WebSockets
Data: PostgreSQL, ETL pipelines, vector search (FAISS, Chroma)
GitHub: https://github.com/ayusrjn
Portfolio: https://ayusrjn.github.io
Email: ayushraanjan@gmail.com