Product Builder at AWS Applied AI | Focused on Agentic and LLM Systems
I build reasoning-driven AI systems, focusing on orchestration, evaluation, governance, and explainability.
I believe the best AI PMs do more than write PRDs. They prototype, test, and iterate.
| Area | Focus |
|---|---|
| 🧩 Agentic Systems | multi-agent orchestration, planning, simulation |
| 📊 Evaluation and Reliability | LLM trust scoring, hallucination detection, reasoning metrics |
| 🧠 Governance and Feedback | bias audits, HITL loops, model transparency |
| 🧰 PM Frameworks | reusable templates and metrics guides for AI teams |
| Repository | Description |
|---|---|
| agentic-reasoning-lab | Multi-agent orchestration and RAG demo with reasoning loop. |
| llm-evaluation-playground | Mock scoring system for LLM accuracy, coherence, and hallucination. |
| feedback-loop-simulator | Simulated human-in-the-loop model feedback cycle. |
| data-lineage-demo | Tracks data flow and governance audit logs in AI systems. |
| model-governance-kit | Simulates bias, fairness, and compliance checks for models. |
| agent-observability-demo | Logs and traces multi-agent interactions in text. |
| sandbox-orchestrator | Modular orchestration of multiple reasoning agents. |
| ai-simulation-framework | Scenario simulation and decision evaluation sandbox. |
| ai-pm-templates | PRD, prompt, and evaluation templates for AI PMs. |
| ai-metrics-guide | Key reasoning and reliability metrics reference for PMs. |
| hallucination-detection-lab | Simple playground for detecting potential hallucinations in model outputs using mock factual checks. |
Python · AWS · LangChain · Bedrock · Kendra · OpenAI · PyTorch · FAISS
GitHub · Markdown · AI Evaluation · Agent Orchestration
📫 linkedin.com/in/iltutishrak
🧰 github.com/iltutishrak