Agentic AI / Data & ML Engineer • MS CS @ NJIT • Open to Work (2026)

I build production-grade data + ML systems and agentic AI copilots that connect to real tools (databases, filings, APIs) and return auditable, citation-first outputs.
My focus is hands-on: reproducible pipelines, reliable evaluation, and clean architecture you can extend.

• Agentic AI systems: single- and multi-agent workflows with tool calling, structured outputs, retries/fallbacks, and safe execution patterns.
• Orchestration: graph-based routing (e.g., LangGraph-style), supervisor/critic patterns, and agent-to-agent task decomposition for complex queries.
• Tooling & integrations: agents that query real systems (SEC/EDGAR, financial statements, macro data, PDFs, GitHub, databases) and return cited, auditable results.
• RAG (production-grade): baseline RAG → agentic RAG → hybrid retrieval (keyword + vector), metadata filtering, chunking strategies, and citation-first responses.
• Memory: short-term conversation memory + long-term task/user memory with clear “when to remember vs. forget” rules.
• Evaluation & observability: automated evals (RAG quality + tool accuracy), tracing, prompt/version tracking, and regression tests for agent behavior.
• LLM app engineering: streaming UX, caching, rate-limit handling, cost/latency optimization, and robust error handling.
• Forecasting + analytics modules: time-series forecasting and KPI analysis exposed as callable tools with explainable outputs.
• Reusable templates: “clone → install → run” starter kits with clean repo structure and reproducible environments.

If you want to scale AI adoption, automate high-impact workflows, or get measurable results with GenAI + ML — let’s connect.