Undergraduate in Computer Science, focused on Artificial Intelligence and Data Science.
I build practical systems around large language models, retrieval, and data pipelines.
Generated nightly from my GitHub activity.
Real-time voice support agent
Low-latency streaming voice agent (16 kHz PCM) that handles order lookup, refunds, and complaints through LLM tool calling. WebSocket-based bidirectional audio, live transcripts, and backend policy logic (refund windows, escalation rules), deployed in production on GCP.
Coding agent system
Multi-step software agent with file operations, command execution, and repository-level search. Persistent memory (todo.md, scratchpad.md, transcript.md), sliding-window context management, and grep-based retrieval to cut token usage and scale to larger repos.
Time-aware data pipelines
Pipelines for sequential and multimodal data with temporal indexing, event ordering, and trend analysis over user-generated content. Reworked backend schema and ingestion to support time-dependent queries and downstream modelling.
Transformer sentiment service
DistilBERT-based service that tracks how narratives evolve over time, with inference pipelines for continuous sentiment monitoring feeding higher-level narrative analysis.
Large Language Models · Retrieval-Augmented Generation · Agent-based systems · Context and memory handling · Data engineering for AI
| Languages | Python · C++ |
| AI / ML | Transformers · LangChain · LlamaIndex |
| Systems | FastAPI · Docker · GCP · WebSockets |
| Data | PostgreSQL · ETL pipelines · FAISS · Chroma |



