OPEN TO COLLABORATION · AI SECURITY · OPTIMIZATION

Building AI systems that are secure, measurable, and useful.

Applied AI/ML engineer focused on cybersecurity, secure tool-using agents, retrieval systems, and operations research. I publish complete, reusable Hugging Face projects spanning data, models, benchmarks, and interactive demos.

Explore my Hugging Face

Every metric and project below links to a public source.

110+ Models 95+ Datasets 100+ Spaces 5+ Curated collections

SELECTED WORK

Flagship project suites

A curated view of the strongest, best-documented systems in a broad public portfolio.

Vulnerability Intelligence

CVE-RiskRank

Operational vulnerability-prioritization corpus joining NVD and CISA KEV data with engineered features and time-aware labels.

  • 245.3K CVE records
  • 40 structured and textual features
  • Temporal train, validation, and test splits
Detection Engineering

SigmaForge

Structured assets for generating, validating, translating, and repairing portable Sigma detection rules across major SIEMs.

  • 9.2K training and evaluation samples
  • Description-to-rule, logs-to-rule, and repair tasks
  • ATT&CK mappings plus Splunk, Elastic, and Sentinel output
Security Operations

SOC Agent Research Suite

Episode-based evaluation and model assets for AI security analysts working across heterogeneous enterprise telemetry.

  • 120 investigation episodes
  • 12 attack-pattern families
  • Sysmon, Zeek, Suricata, Wazuh, cloud, and Kubernetes logs
Decision Intelligence

Operations Research AI

Models and interactive systems for mathematical formulation, solver selection, code generation, scheduling, routing, and supply-chain decisions.

  • Two 1.54B MiniZinc generation and repair models
  • Pyomo, MiniZinc, OR-Tools, CP-SAT, and MILP workflows
  • Execution-based benchmarks and interactive solver demos

EXPERTISE

Technical focus

Technologies evidenced across public model cards, dataset cards, and deployed Spaces.

01

Secure AI & Cybersecurity

Agent and MCP security, prompt injection, SOC workflows, SIEM, MITRE ATT&CK, Sigma, CVE/CWE, malware and phishing analysis.

02

ML & LLM Engineering

PyTorch, Transformers, Sentence Transformers, SetFit, PEFT, QLoRA, TRL, scikit-learn, LightGBM, embedding and reranking.

03

Data & Evaluation

Hugging Face Datasets, Parquet, Polars, schema design, hard negatives, temporal splits, reproducible benchmarks and metrics.

04

Optimization & OR

Mathematical programming, Pyomo, MiniZinc, OR-Tools, CP-SAT, MILP, scheduling, routing, solver selection and supply chains.

05

Product & Deployment

Gradio, Hugging Face Spaces and ZeroGPU, Docker, FastAPI, interactive evaluation consoles, model and dataset documentation.

06

Enterprise & Bilingual AI

English/Persian document AI, OCR, RAG, structured extraction, and applied systems for security, logistics, retail, and operations.

ENGINEERING APPROACH

From raw problem to an evaluable product.

I structure open-source work as connected assets rather than isolated demos. This makes every project easier to inspect, reproduce, compare, and reuse.

  1. 01 Dataset Validated schemas and documented provenance
  2. 02 Baseline & model Practical reference points and fine-tuned artifacts
  3. 03 Benchmark Task-specific, reproducible evaluation
  4. 04 Interactive Space Hands-on inspection and transparent outputs
  5. 05 Collection A navigable, end-to-end project narrative

COLLABORATION

Let’s build secure and measurable AI.

Explore the repositories, open a Community discussion on a relevant project, or follow the account for new releases.