Main components of the project include the data ingestion and retrieval pipeline. Used public test data, which can be used for Motorola's private internal documents.
Developed internal RAG evaluation dashboard to benchmark and visualize retrieval-augmented generation pipelines.
Enabled engineering teams to assess retrieval quality, context grounding, and hallucination risk.
Integrated embedding models via SentenceTransformers for encoding document chunks into vector representations.
Implemented cosine similarity for matching queries to top-k relevant chunks for LLM processing.
Established metrics for diagnosing failure modes in RAG systems, including faithfulness and relevance.
Computed token-level metrics to evaluate context utilization within answers effectively.
Supported diverse applications such as internal knowledge search and engineering documentation retrieval.
Finally, presented the product to company mentor and faculty advisor for approval.
Motorola MCP
Motorola Solutions
Richardson, Texas
08.2025 - 01.2026
MCP Context Assembly: an MCP server (SDK-based) + client that (a) proves protocol mastery via bare-metal JSON-RPC, (b) exposes tools/resources/prompts, and (c) builds session-aware memory_context bundles with token budgeting for stateful conversations.
Provided support and guidance to colleagues to maintain a collaborative work environment.
Conducted comprehensive research and data analysis to support strategic planning and informed decision-making.
Student Intern
AT&T
Plano, TX
06.2020 - 08.2020
Used JavaScript and HTML to create custom websites for the company
Collaborated with other students to learn more about front end using package managers, build tools and Version Control
Launched multiple EC2 instances for personal projects and created billing alarms by using CloudWatch
Education
Bachelor of Science - Computer Science
The University of Texas At Dallas
Richardson
12-2025
Skills
C/C and Java
Python
Bash scripting
SQL database management
AWS services and EC2
Amazon S3 and VPC
Elastic Load Balancing (ELB)
CloudWatch monitoring
Identity and Access Management (IAM)
Unix and Linux systems
Windows operating system
Cloud computing
Software development
Data analysis
Data evaluation
Projects:
AI Agent Development Framework (Google Cloud + ADK) - Built and deployed modular AI agents using Google AI Studio API and Agent Development Kit (ADK), integrating secure API key management and cloud-based project configuration. Implemented multiple agent architectures including basic conversational agents, tool-augmented agents, multi-agent orchestration, and stateful session-based systems which was used in my experience for Motorola Solutions
Developed structured output pipelines using Pydantic schemas to ensure consistent model responses, implemented persistent storage for cross-session memory, and designed sequential and parallel agent workflows for improved performance. Leveraged LiteLLM for flexible model abstraction and callbacks for real-time monitoring of agent behaviors.
Demonstrated advanced agent patterns including multi-agent collaboration, iterative loop refinement, and stateful task orchestration for scalable AI application development.
Alexa Skill, AWS Lambda, Amazon Polly, Used lambda functions with AWS Serverless Application Repository, Deployed Alexa skill to Lambda and created custom Mp3 audio source