Ebrahim Azarisooreh

Senior Software Engineer | AI & Systems Architect

Email GitHub

About

Senior Software Engineer with 10+ years of experience optimizing enterprise-scale systems, delivering high-performance backend solutions, and architecting agentic LLM infrastructure that scales engineering productivity across teams. Proven track record of reducing system response times by 70%, cutting build times by 93%, and processing billions of data combinations in milliseconds.

My expertise spans backend development, scalable modular systems, and Gen AI integration for production engineering environments. I specialize in custom MCP tooling, agentic workflow orchestration, knowledge management systems with LLMs, and enterprise GitHub Copilot integration. From low-level systems programming in C and Rust to distributed architectures using AWS, Kafka, and message-bus systems, I'm passionate about building elegant, efficient solutions to complex technical challenges.

Experience

Senior Software Development Engineer I

Travelport January 2020 – Present | Denver, CO
  • Architected agentic development infrastructure across team codebase using enterprise GitHub Copilot (OpenAI, Anthropic, Google models), defining custom agent configurations for code conventions, parallel build validation via make, and domain-specific modular coding patterns — scaling AI-assisted development across a multi-service team
  • Built custom MCP tooling that performs semantic diffs on hierarchical transaction structures, producing structured data summaries — leveraged both by engineers for human review and by tester agents for automated validation of agent-driven changes against baseline runs
  • Designed opt-in MCP tool providing persistent session state that captures agent tool invocations as replayable instructions, enabling on-demand generation of reusable scripts from agent-collaborative workflows worth preserving
  • Authored institutional agent skills automating test plan generation, compliance paperwork, and environment-specific application behavior management — eliminating recurring toil across engineering workflows
  • Implemented progressive disclosure of agent instructions via nested configuration files throughout the project hierarchy, surfacing context on an as-needed basis to optimize token consumption while preserving comprehensive coverage across the codebase
  • Engineered agentic knowledge management system over an Obsidian vault: ingest-time curation, indexed article retrieval, and custom skills for query (via qmd with BM25, vector similarity, and local LLM re-ranking) and write-back of new insights — validated on personal AGI research project with working runs being adapted for institutional deployment
  • Reduced fare search system response times by 70% in the most challenging scenarios (10 to 3 seconds) while cutting CPU utilization in half, directly improving customer experience and reducing AWS infrastructure costs
  • Architected and implemented multi-destination search engine processing billions of flight combinations in under 1 second, optimizing for customer convenience with minimal connections and lowest fares, resolving multiple critical customer issues
  • Achieved 93% reduction in application build times (10 minutes to 40 seconds), dramatically accelerating development cycles and team productivity across multiple engineering teams
  • Led AWS cloud migration initiative, implementing automated monitoring systems with self-healing capabilities for critical production servers, improving system uptime to 99.9%

Software Engineer

Kyndi January 2017 – June 2019 | San Mateo, CA
  • Parallelized NLP computations achieving performance improvements in natural language understanding pipeline, enabling real-time text processing
  • Designed distributed message-bus architecture using Hadoop, Kafka, and RabbitMQ for real-time data processing at scale
  • Built extractive summarization pipeline combining word embeddings (Word2Vec, FastText), lexical-semantic relations (Wordnet), and graph-based ranking (TextRank) over Stanford CoreNLP parses, improving content understanding accuracy
  • Established comprehensive automated testing framework including Stanford Q&A benchmarks and robust unit test coverage for compiler libraries
  • Enhanced API design and implementation, improving system modularity and enabling easier integration for downstream services

Skills

Languages

C, C++, Java, C#, Rust, Python, Prolog

LLM & Agentic Systems

GitHub Copilot (Enterprise), Model Context Protocol (MCP), Custom Agent Design, Prompt Engineering, Agentic Workflow Orchestration, RAG (Hybrid Retrieval, Reranking), Knowledge Base Curation

Cloud & Infrastructure

AWS (EC2, Lambda, CloudWatch), Linux Administration, Docker, CI/CD Pipelines

Data Systems

Hadoop, Kafka, RabbitMQ, Message Bus Architecture, Distributed Systems

Education

Bachelor of Science in Computer Science

University of Colorado Boulder

Bachelor of Arts in Psychology

University of California, Riverside