Skills
Not a list of things I've heard of. A list of things I've used to build something real.
---
Languages
Python — Primary language. Production IAM pipelines, ML model training and evaluation, REST APIs with FastAPI, detection pipelines, data processing at scale (2.5M+ records), scripting cloud automation. This is where I do most of my serious work.
C++ — Built a custom static site generator from scratch: markdown parser, Jupyter notebook converter, incremental build cache with file hashing, SQLite integration, modular architecture across 15+ compilation units. C++17 with CMake and Make.
Java — Android development. Built the mobile client for a real-time face detection app with asynchronous REST calls and camera overlay rendering.
SQL — PostgreSQL for production applications (digital library platform, 5K+ users). SQLite for embedded database in C++ projects. Query optimization matters when you have concurrent users during exam season.
JavaScript — Client-side search and filtering logic for static sites. React for frontend components in the DSA coaching platform.
Bash/Shell — Automation scripts, deployment pipelines, Linux system administration, inference stack setup.
---
Machine Learning & AI
Architectures I've trained:
- Temporal Convolutional Networks (TCN) — sequential flow analysis for DDoS detection
- LSTM Autoencoders — anomaly detection on network traffic (UNSW-NB15)
- Hybrid LSTM-CNN — multi-class attack classification across 9 attack categories
- Haar Cascade CNNs — real-time face detection via OpenCV
What I understand beyond the model:
- Feature selection as a core engineering discipline, not a preprocessing afterthought
- Metaheuristic optimization: Sine Cosine Algorithm, Particle Swarm Optimization — used in production research to solve feature selection as a search problem
- Dimensionality reduction under accuracy constraints (317 → 32 features, 90% reduction, minimal accuracy loss)
- False positive rate as the production metric that matters in security contexts — not just accuracy or F1
Frameworks: TensorFlow · Keras · scikit-learn · Pandas · NumPy
LLM Integration: Gemini API — building natural language alert explanation pipelines for IDS output. Prompt engineering for structured security event summarization.
Local Inference: llama.cpp · quantized model deployment (GGUF/GGML) · throughput and latency benchmarking under concurrent load
---
Security Engineering
Identity & Access Management:
- Automated IAM lifecycle (provision → modify → deprovision) across Ivanti, Google Workspace, and Active Directory
- RBAC implementation for production systems with 5,000+ users
- OAuth 2.0 — designed and implemented for a production digital library platform
- Session management, privilege escalation detection, access anomaly flagging
Cloud Security (AWS):
- IAM policy refactoring to NIST 800-53 AC controls and CIS AWS Foundations Benchmarks v1.4
- Detection pipelines for configuration drift, privilege escalation, access anomalies
- AWS Lambda for event-driven detection logic
- Elasticsearch for log aggregation and anomaly search
Network Security:
- Intrusion Detection System design — academic research and capstone project
- Protocol analysis: TCP/IP, UDP, QUIC, BGP fundamentals
- Traffic flow feature engineering: packet size distributions, inter-arrival times, TCP flag patterns, flow duration statistics
- Wireshark for packet analysis
- Network architecture design — studied in depth at Penn State
Security Concepts I Can Apply (Not Just Define):
- MITRE ATT&CK framework — mapping detection rules to TTP coverage
- NIST 800-53 control families — AC, AU, SI in particular
- CIS Benchmarks — AWS Foundations, practical hardening
- Secure network architecture — segmentation, DMZ design, access control layers
- Binary analysis fundamentals — reverse engineering coursework at Penn State
---
Cloud & Infrastructure
AWS: IAM · Lambda · CloudWatch · S3 · foundational compute and networking
Docker: Containerized microservices for the digital library platform. Container-based deployment for reproducibility and horizontal scaling.
GitHub Actions: CI/CD pipeline for automated site generation and deployment to GitHub Pages.
Linux: Primary development environment. System administration, process management, network configuration, inference stack setup on local hardware.
FastAPI: REST API backends — face detection service, DSA coaching platform backend. Async request handling, endpoint documentation, dependency injection.
Elasticsearch: Log aggregation and search for cloud security detection pipelines at Invisbl.
---
Tools & Platforms
| Category | Tools |
|---|---|
| ML/Data | TensorFlow, scikit-learn, Pandas, NumPy, Jupyter |
| Monitoring | Elasticsearch, CloudWatch |
| Identity | Ivanti, Google Workspace Admin, Active Directory |
| Databases | PostgreSQL, SQLite, MySQL |
| Version Control | Git, GitHub, GitLab |
| Containers | Docker |
| API | FastAPI, REST, Postman |
| Networking | Wireshark, QUIC stack |
| Local AI | llama.cpp, GGUF quantization |
| Build Systems | CMake, Make, GitHub Actions |
---
Certifications
CompTIA Security+ Cryptography, PKI, network security protocols, threat management, vulnerability assessment, identity and access management, risk management. Got this to formalize the fundamentals I'd been applying in security engineering roles.
Cisco CCNA Routing protocols (OSPF, BGP behavior, EIGRP), switching, VLANs, subnetting, network access control, WAN technologies. The CCNA forced me to understand what actually happens on the wire — not just the abstractions.
---
Education
MS in Cybersecurity — Penn State University GPA: 3.8 · Graduating May 2026 Coursework: Network security, secure network architecture, binary analysis and reverse engineering, applied cryptography, risk management, cloud security
Integrated MSc in Information Technology (5-year program) Coursework: Machine learning, data structures and algorithms, database systems, computer networks, operating systems, distributed systems
---
What I'm Learning Right Now
- P99 tail latency behavior in QUIC queue scheduling (ongoing simulation work)
- Local LLM inference optimization — quantization tradeoffs, memory bandwidth limits, concurrent request handling
- Designing Data-Intensive Applications — reliability, consensus, and replication chapters map directly onto security architecture thinking
---
If something here overlaps with what you're building, let's talk.