"Programming is thinking, not typing!"

• AI Software Engineer @ Chewy, building production GenAI and agentic AI systems for Customer Care.
• Research Volunteer, weekends--unpaid, with the AGAI group.
• Research Thesis: Blindfolded Spider-Man Optimization A single-point metaheuristic for continuous and discrete search spaces.
• Published work on LLM-based course recommendation systems in IEEE Access.
• Working at the intersection of AI / ML / LLMs / Optimization / Agentic Systems.
• Focused on building production-grade, scalable, observable, and reliable AI systems.
• Open to collaborating on practical AI, GenAI, RAG, agentic AI, and applied research projects.

• Research: Optimization Algorithms, Metaheuristics, LLM Hallucination Detection and Mitigation, AI Reliability, Security in Language-Model-Based Systems.
• Languages: Python, Java, C++
• AI / ML: PyTorch, Scikit-learn, MLflow, Hugging Face, Deep Learning, Neural Networks, NLP, LangChain, LlamaIndex, LangGraph
• GenAI: RAG, Agentic AI, Multi-Agent Systems, Graph RAG, MCP, AI Memory, Prompt Engineering, Prompt-as-Code, LLM Observability, GenAI Evaluations
• Data / Infra: AWS, AWS Glue, Airflow, SageMaker, Docker, CI/CD, MLOps, AIOps, Vector Databases, OpenSearch

• Central hub for AI & ML tutorials, cookbooks, and practical implementation guides.
• Covers RAG, Agentic AI, LLMs, Fine-tuning, Graph RAG, and real-world GenAI systems.
• Includes hands-on Jupyter notebooks with end-to-end examples.
• Built for students, engineers, and AI builders who want practical, code-first learning.

• PyPI: https://pypi.org/project/instructvault/
• GitHub: https://github.com/05satyam/instruct_vault

pip install instructvault

InstructVault is a Git-first prompt-as-code platform for managing prompts as versioned, auditable artifacts.

• Version prompts like code.
• Track prompt updates through Git.
• Support CI-based validation and deterministic evaluations.
• Use release tagging for safer prompt deployment.
• Load prompts locally without adding runtime latency or vendor lock-in.

• PyPI: https://pypi.org/project/nonconvexoptimzationfunclib/
• GitHub: https://github.com/05satyam/nonconvexoptimizationfunclib

A Python library of standard non-convex benchmark functions for evaluating and comparing optimization algorithms.

pip install nonconvexoptimzationfunclib

Highlights:

• Includes standard benchmark functions for optimization research.
• Supports continuous and discrete optimization experiments.
• Useful for testing metaheuristics and comparing optimization algorithms.
• Built from my research interest in optimization and search methods.

Blindfolded Spider-Man Optimization is a single-point metaheuristic designed for continuous and discrete search spaces.

Research focus:

• Metaheuristic optimization
• Continuous and discrete search
• Unbounded knapsack problems
• Higher-dimensional benchmark testing
• Optimization with memory-inspired movement patterns

Research Volunteer with the AGAI Research Group under guidance of Dr. Farag.

Project focus:

• Designing a multi-agent framework for hallucination filtering and disinformation mitigation.
• Studying why LLMs hallucinate, including context loss, retrieval bias, and reasoning drift.
• Exploring agent-based verification, evidence retrieval, and response validation.
• Working toward more reliable, explainable, and trustworthy generative AI systems.

IEEE Access

• Published a study on how LLMs are changing course recommendation systems in education.
• Covered recommendation quality, explainability, personalization, and the role of LLM reasoning.

arXiv Preprint

• Proposed a single-point metaheuristic algorithm for continuous and discrete optimization.
• Evaluated the approach across benchmark functions and unbounded knapsack experiments.
• Explored memory-inspired movement patterns for search in higher-dimensional spaces.

• Delivered a hands-on workshop on low-code AI tools.
• Guided attendees through building practical AI applications.
• Focused on taking ideas from prototype toward production-ready applications.

• Presented work on course recommendation systems using educational data mining and learning analytics.
• Explored domain-clustered SVD++ and RAG-based LLM reasoning for explainable course recommendations.
• Work connected to DOTAIZ, an AI-powered micro-learning and recommendation platform.

• Presented work from an NSF-funded DEI research project on faculty hiring trends.
• Contributed to a centralized hiring analytics dashboard using CIP codes and institutional demographic data.
• Supported data wrangling and visual analysis across 23 California State University campuses.

Topic: Exploring Naive RAG and Agentic RAG with Hands-On

• Led a live webinar on RAG workflows, evaluation strategies, and advanced AI pipelines.
• Shared practical guidance on AI engineering interviews and production AI skills.
• Session attended by 70+ students and practitioners globally.

Topic: Building AI Applications with RAG and Agentic RAG

• Delivered live sessions on foundational and advanced RAG concepts.
• Shared code demos for building custom agentic RAG systems.
• Discussed fine-tuning and domain-specific GenAI application design.
• Sessions attended by 80+ students and practitioners globally.

Topic: Fundamentals of MCP and A2A for Agentic AI Systems

• Led a hands-on webinar on MCP and A2A for agentic AI systems.
• Discussed moving GenAI applications from PoC to production.
• Covered the role of evaluation frameworks in reliable AI applications.
• Session attended by 70+ students and practitioners globally.

• My favorite sitcom is The Big Bang Theory 🧠
• Yes, One Piece truly exists! ☠️🏴‍☠️

⭐ If you find my work useful, consider starring the repositories!