Ahmed.currently: open to work ✦
I build production-grade LLM systems, multi-agent architectures, and agentic workflows that ship — not just demo. 6+ years taking AI from whiteboard to deployed infrastructure at companies like Vanguard, TXU Energy, and Yellow AI.
AI systems
via ML migration
speed-up
in production
Multi-agent GCP knowledge assistant. A natural-language query hits a LangGraph orchestrator which routes to specialist agents — RAG retrieval, BigQuery analytics, sandboxed Python execution — then synthesizes one coherent, cited response. Production-deployed on Cloud Run with full CI/CD, Langfuse observability, and RAG evals.
Real-time collaborative task management with append-only event sourcing, WebSocket delta sync, and AI task suggestions powered by OpenRouter and Anthropic in parallel. Built in Go + Next.js with optimistic locking, undo/redo, presence awareness, and Redis pub/sub for multi-instance fan-out.
- Built LangGraph multi-agent graph with Postgres checkpointing + Redis session state
- Delivered MCP server/client integrations over stdio and StreamableHTTP
- Established RAG evaluation loop with Langfuse tracing for regression tracking
- Embedded with execs: scoped POC → production handoff with full runbooks
- Led GenAI initiatives from internal demo to accepted enterprise rollout
- Deployed LLMs via AWS Bedrock integrated with DynamoDB at scale
- A/B tested multi-agent MCP architectures with segmented employee groups
- Coordinated rollout across 5 stakeholder groups
- Improved LLM code generation accuracy across Python, Java, JavaScript, Go
- Applied chain-of-thought and ReAct prompting — reduced hallucinations measurably
- Generated training data supporting RLHF reward model optimization
- Reduced licensing costs 70% migrating SAS → Python + AWS Lambda/Glue
- Cut analytics processing from 20+ hrs to under 2 hrs via cloud-native ETL
- Built churn prediction model with strong accuracy on behavioral + temporal data
- Improved customer verification to 98% accuracy with deep learning ID scan
- Increased client revenue 20% via automated analytics microservices
- Scaled enterprise chatbot deployment across ITSM and customer service environments
HarvardX Data Science Professional
build.
Open to senior AI engineering and Forward Deployed Engineer roles at companies shipping real AI products. I move fast, own systems end-to-end, and care about what gets deployed — not just what gets demoed.