Houston, TX
📧 [email protected]
📱 +1 713-853-7974
🌐 Ibrahimsaleem.com | Portfolio | LinkedIn | GitHub

University of Houston — Master of Science in Cybersecurity
Houston, TX | Expected August 2026
GPA: 4.0/4.0
Relevant Coursework: Network Security, Secure Enterprise Computing, Cryptography, Data Analysis for Cybersecurity, Risk Management

Rajiv Gandhi Proudyogiki Vishwavidyalaya — Bachelor of Technology in Computer Science Engineering
July 2023
Relevant Coursework: Python, Data Structures & Algorithms, DBMS, Cloud Computing, Operating Systems, Networking Protocols (TCP/IP, DNS, HTTP, SSL/TLS), Cryptography, Machine Learning

NOV Inc (National Oilwell Varco) | Houston, TX (On-site)
June 2025 – Present

• Led the design and deployment of GenAI-based automation pipelines, reducing multi-day manual workflows to under 5 minutes.
• Improved automation accuracy from 92% to 98% by incorporating adaptive prompting, context-aware logic, and feedback-driven retry mechanisms. Research Paper
• Developed and productionized a Modular Command & Processing (MCP) server to orchestrate autonomous AI agents, introducing self-healing capabilities.
• Built an end-to-end OCR and LLM pipeline for PDF data extraction, leveraging RAG architectures and integrating with Streamlit for real-time validation.
• Integrated AI workflows into Databricks and AWS, automating data ingestion, inference, logging, and evaluation.
• Enhanced internal AI tooling by deploying Streamlit-based GenAI dashboards for non-technical users.
• Logged version-controlled results for continuous evaluation of LLM, OCR, and pipeline performance.

University of Houston | Houston, TX (On-site)
Sep 2024 – May 2025
Paper | PentestThinkingMCP Repo

• Led the design and implementation of LIMA as first author — a modular, LLM-driven penetration testing framework combining GPT-4o/Claude with an autonomous MCP-based system.
• Developed the PentestThinkingMCP server, integrating advanced reasoning agents with tools like Nmap, Metasploit, enum4linux, and BrowserMCP.
• Designed and executed the first quantitative benchmark comparing AI and human performance on HackTheBox machines; Claude 3.5 outperformed expert pentesters on multiple boxes with less than $0.05 run cost.
• Contributed novel insights into LLM reasoning weaknesses (e.g., GUI/tooling limitations) and adaptive failure points in real-world cybersecurity environments.
• Implemented a secure automotive IoT pairing mechanism using unencrypted TPMS signals and DSP techniques in Python and MATLAB.
• Synthesized findings from over 40 peer-reviewed papers to align research with current academic and industry trends.

Nagarro Software Pvt. Ltd.
March 2023 – February 2024

• Designed and developed scalable backend solutions using C#, .NET Core, and SQL Server, optimizing database queries and building secure REST APIs.
• Developed interactive Angular-based UI dashboards and automated data analytics pipelines using Python and Pandas.

State Cyber Cell MP Police
July 2022 – December 2022

• Conducted security assessments on critical infrastructure systems and delivered cybersecurity awareness training to 500+ students and professionals.

• CyberPath AI – Personalized Learning Roadmaps
  Flask, GenAI, Python
  AI-driven web app that generates custom, interactive roadmaps for Cybersecurity, AI Engineering, and Data Science. Users click any topic to receive an on-demand explanation crafted by Gemini 2.0, dynamically tailored to their background and experience.

• PentestThinkingMCP – Autonomous Pen-Testing Server
  MCP, Beam Search, MCTS, AI Agent
  Designed an MCP Server which helps LLMs reason, plan, and execute complete attack chains using Beam Search + Monte Carlo Tree Search. Fully compromised HTB "Lame" in 3 min end-to-end for ~$0.03 in LLM cost.

• LocalRAGAgent – Offline Retrieval-Augmented QA
  Python, LangChain, FastAPI, Chroma, Ollama
  Built a 100% local RAG pipeline that extracts text OCR from PDFs, embeds chunks into ChromaDB, retrieves context, and answers queries with Llama-3 via Ollama, delivering <300 ms latency on consumer hardware.

• Secure Offline AI Assistant for Network Operators
  Huggingface, Flask, LLM
  Fine-tuned a Llama 3.2 3b model via Unsloth on domain-specific logs, delivering on-prem NLP troubleshooting assistance and reducing mean-time-to-resolution by 30%.

• Advanced Network Performance Monitoring System
  Python, ML, SDN (Ryu)
  Integrated ML-based DDoS detection (98% accuracy) with SDN controllers; automated real-time mitigation using MITRE ATT&CK tactics, improving network resilience.

• Splunk-Based Detection & Response Automation
  Python, Splunk
  Built security dashboards and AbuseIPDB-powered reputation scoring that cut manual triage workload by 45%.

• Cybersecurity Awareness Platform
  Python, JavaScript, Flask
  Launched an interactive training suite featuring 30+ Python-built security labs; increased learner completion rates by 40%.

• Languages & Scripting: Python, JavaScript/TypeScript, C#, SQL, Bash/Shell, PowerShell
• Frameworks & Libraries: Node.js, React, Angular, Flask, FastAPI, .NET Core, Streamlit, LangChain, PyTorch, TensorFlow, scikit-learn, Hugging Face, Llama/Ollama
• Cloud, DevOps & MLOps: AWS (S3, Lambda, SageMaker, IAM, VPC), Azure, GCP, Docker, Kubernetes, Terraform, Jenkins, GitHub Actions, Databricks, CI/CD, Bandit, SonarQube
• Data & ML Engineering: Pandas, NumPy, ChromaDB, ETL/ELT pipelines, data warehousing, real-time analytics
• Generative AI: LLM fine-tuning/distillation (GPT-4o, Claude, Llama-3), prompt engineering, RAG, autonomous agent orchestration (MCP, LangGraph), RLHF, synthetic data generation, model serving/evaluation, Explainable AI
• Databases: SQL Server, MySQL, PostgreSQL
• Cybersecurity & Networking: Secure coding, Nmap, Metasploit, enum4linux, SDN (Ryu), Splunk, AbuseIPDB, MCP security agents, risk assessment, incident response
• Software Engineering: System design, distributed systems, RESTful APIs, TDD, debugging, automation pipelines, real-time data visualization
• Professional: Research, Agile/Scrum, project management, technical communication
• Certifications: AWS Certified Security – Specialty, CompTIA Security+, Microsoft Azure Security Fundamentals, Python for Networking Security, plus additional AI/ML/MLOps credentials

• Saleem, M. I. (First Author), Prabhakar, S. S., Harsha, A., Nagabhushan, D., Conklin, W. A., Lee, K. I., Banerjee, T.
  "LIMA: Leveraging Large Language Models and MCP Servers for Initial Machine Access."
  Proc. IEEE Intl. Conf. on Future Machine Learning and Data Science (FMLDS), 2025 — accepted.
• NOV GenAI Team, Saleem, M. I. (Second Author).
  "Self-Improving GenAI Agent for Fully Automated Report Parsing in Enterprise Environments."
  Manuscript in preparation, 2025 — work conducted as GenAI & Data Science Intern, NOV Inc.

"Code is like humor. When you have to explain it, it's bad." – Cory House