Software Engineer
Software engineer. I build AI powered infrastructure, data pipelines, and agentic workflows that actually ship.
Production RAG engines. MCP agent tooling. Multi model LLM orchestration. Concurrent systems built to handle real data at real scale.
About
I'm a software engineer. Backend architecture, AI systems, cloud automation. I write Python pipelines, vector retrieval systems, and API integrations that take messy operational work and turn it into software you can run on a schedule and trust.
Featured Projects
View all 30 →Stateful Multi-Agent Compliance Orchestrator
Stateful LangGraph workflow on ECS Fargate with DynamoDB checkpointing, S3-backed Claim Check state, idempotent pause/resume, and human-in-the-loop escalation.
- →DynamoDB checkpointing and S3 Claim Check state support pause/resume.
- →Idempotency guards retries; a bounded reviewer routes uncertain evidence to HITL.
Production RAG Infrastructure on AWS
Python retrieval infrastructure on OpenSearch and Bedrock with hybrid search, multi-tenant isolation, and freshness-aware indexing.
- →Namespace isolation serves 15+ AI pipelines from one retrieval layer.
- →Timestamp checks skip unchanged documents and scope stale-vector re-indexing.
Compliance Evidence Integration Platform
Python Lambda and Terraform integration platform with 23 scheduled adapters across AWS, GCP, and GitHub; shared auth, API clients, and test coverage.
- →Source adapters and output targets support 23 scheduled integrations without pipeline forks.
- →One auth interface covers OAuth, GitHub App, and Workload Identity Federation; 85 tests cover the platform.
Artemis: FedRAMP Security Graph Engine
Python and DSPy security graph linking FedRAMP controls to MITRE ATT&CK and D3FEND, with interactive coverage analysis and OSCAL output.
- →Maps 323 FedRAMP controls to MITRE ATT&CK and D3FEND coverage.
- →Exports OSCAL incident-response artifacts and interactive graph views.
Actuarial Cyber Risk Engine
Python Monte Carlo service that estimates Expected Loss, VaR, and Tail VaR from compliance posture, deployed on Lambda and API Gateway with CDK.
- →Runs 10,000 simulated years with seeded, reproducible risk curves.
- →Serves API-key-protected results from Lambda and API Gateway.
Full-Stack AI Agent Chatbot
Next.js chat UI and FastAPI backend using GPT-4o tool calling to query live Azure resource inventory.
- →FastAPI queries Azure Resource Management with server-side credentials.
- →Next.js streams responses and tool-call state from a bounded agent loop.
More Projects
View all 30 →AI Driven Issue Tracking and Analytics Pipeline
An eight stage AI pipeline. It maps FedRAMP 20x controls to a client's actual tech stack, finds the gaps against live Vanta test data, writes the remediation plan, and uploads the whole Epic, Task, and Subtask hierarchy to Jira.
Two-Stage RAG Reranking Pipeline
A FastAPI RAG service that improves answer precision with a two-stage retrieve-then-rerank design: dense vector search fetches candidates, a cross-encoder re-scores them, and only the top results are passed to the LLM — with citations and a strict no-context guard.
Jira MCP Server for AI Agent Workflows
An MCP server that exposes Jira operations as tools the LLM can actually call. Search issues, create them, update them, transition them, drop comments. Not just suggest what to do. Do it.
Cloud Data Normalization Pipeline
A data pipeline that takes raw cloud inventory exports and Tenable scan data and turns it into a FedRAMP Appendix M workbook you can actually submit.
Prompt Engineering That Stops LLMs from Lying in Evidence
A layered prompt and validation architecture that keeps LLMs from hallucinating in compliance evidence work. Structured inputs, hard constraint gates, four phases of validation.
System Architecture & Design
A high level architecture diagram of the compliance automation platform. Orchestration layer, DI, strategy based backend execution, logic pipelines, adapter protocols, and external provider integrations, all laid out.
Skills
Core
Tools
Contact
Looking to build AI powered systems that solve real operational problems. If that's the kind of work your team does, get in touch.