OBJECTIVE: Establish real-time AI financial intelligence feed.
OUTCOME: Engineered event-driven pipeline. Ingests news via Redpanda, sentiment analysis via local Llama 3. Sub-25ms latency achieved.
REDPANDA
LLAMA 3
DOCKER
OBJECTIVE: Secure offline document search (RAG).
OUTCOME: Privacy-first architecture eliminating cloud data leaks. Orchestrated Qdrant vector DB with local LLMs.
LANGCHAIN
QDRANT
FASTAPI
OBJECTIVE: Automate ML lifecycle for churn prediction.
OUTCOME: Containerized training pipeline integrated with MLflow tracking and GitHub Actions CI/CD.
MLFLOW
DOCKER
CI/CD