"Moving without traditional guardrails, I treat advanced AI and LLMs as core engineering partners. By fusing rapid execution with local AI orchestration, I turn raw technical curiosity into production-grade systems entirely from scratch."

A terminal-first, local-first AI developer workspace engineered for repository-scale code intelligence, task classification, hybrid retrieval, and evidence-driven patch verification.

• Developer Workflow Pipeline: Developer Intent ➔ Task Classification ➔ Model Router ➔ Hybrid RAG ➔ Patch Generation ➔ Evidence Verification
• Deterministic Task & Model Routing: Built around a single authoritative routing boundary (TaskClassifier ➔ ModelRouter ➔ LLMFactory) to dynamically evaluate task complexity and map prompts to the optimal provider (Ollama / Local LLMs).
• Hybrid Retrieval Architecture: Integrates Dense Vector Embeddings (all-MiniLM-L6-v2 via FAISS) with BM25 Lexical Search for persistent, hash-based incremental indexing of local codebases.
• Safe Patching & Provenance: Executes safe diff validation with path-traversal protection, conflict handling, and inspectable provenance tracing context directly back to retrieved sources.
• Multi-Interface Access: Features unified backend architecture accessible via FastAPI REST endpoints, Textual/Rich CLI/TUI interfaces, and MCP (Model Context Protocol) tools.

An advanced, AI-driven editing suite designed for seamless media manipulation, text-to-speech, speech-to-text workflows, and generative media execution.

• Frontend Architecture: Cross-platform Flutter (Dart) mobile UI utilizing custom Montserrat typography systems and Lottie animation bundles.
• Backend Core: High-performance Python (Flask) engine managing secure user authentication and asset processing.
• REST API Endpoints:
  • GET /api/usage — Real-time AI token tracking and consumption reporting.
  • POST /api/toggle-feature — Dynamic switching of active AI editing capabilities.
  • POST /api/login — Secure token-based user authentication.

An asynchronous dual-engine task execution pool built in Python featuring automatic retry backoff, fault isolation, and automated GitHub Actions CI testing.

• Dual-Engine Architecture: Seamlessly toggles between ThreadPoolExecutor for network/disk I/O and ProcessPoolExecutor to bypass the GIL for CPU-bound multi-core computation.
• Resilience & Fault Isolation: Implements exponential/linear backoff logic (submit_with_retry) with isolated fault handling to aggregate execution states without crashing worker threads.
• Zero External Dependencies: Native Python implementation fully tested using pytest and continuous integration pipelines.

Production-ready backend frameworks featuring automated script execution, rigorous endpoint verification, and cryptographic hash verification using SHA-256 protocols.

🌐 Connect & Collaborate