AlphaBrain Core is a production-grade financial forecasting - An AI engineering system built on AWS and open-source technologies.
To build a full-stack, reproducible, cloud-native AI system that:
- Ingests and processes financial & alternative data.
- Trains and registers baseline models via MLflow.
- Serves inference through containerized FastAPI services on AWS ECS/Fargate.
- Integrates LangChain, LangGraph, and CrewAI for agentic reasoning workflows.
- Implements full observability, automation, and reproducibility using open-source tools.
Infrastructure: AWS (S3, ECS/Fargate, RDS, IAM, Terraform, CloudWatch, Secrets Manager)
Pipelines: Airflow, MLflow, DVC, Terraform
Datastores: S3 (data lake), Postgres (metadata), Milvus or pgvector (embeddings)
API & Serving: FastAPI, Docker, ECS
LLM & Agents: LangChain, LangGraph, CrewAI, OpenAI Evals, Guardrails
Monitoring: Prometheus, Grafana, CloudWatch
A deployed, reproducible system that:
- Ingests daily stock & news data into S3.
- Trains a baseline hybrid model and registers it in MLflow.
- Serves inference via a FastAPI endpoint on AWS ECS.
- Logs all metrics and traces through Prometheus/Grafana.
- Includes a simple LangChain RAG pipeline integrated with the inference API.
This project is for research and educational purposes only.
It does not provide financial advice, execute trades, or make investment recommendations.
Any models, signals, or outputs are intended solely for demonstrating AI engineering practices.
All third-party datasets, models, and APIs must comply with their respective licenses.
Use responsibly and ethically in alignment with data-provider terms.
Achyut Kulkarni — as the primary developer and architect of AlphaBrain Core.
This repository represents your journey toward staff-level AI engineering mastery.# alphabrain-core
AI Engineering for Financial Forecasting