AI + Full Stack Engineer. I build developer tools, production web apps, and IoT systems.
Right now I work mostly on agent infrastructure and AI dev tooling, with Django, FastAPI, and React on the product side and nRF hardware on the edge side.
mohib=Developer(
focus=["AI Dev Tooling", "Agent Security", "Django/FastAPI", "React/Next.js", "IoT/Embedded"],
currently_building="MCP protocol tooling + AI-agent security & evaluation",
caffeinated=True,
)AI/ML: PyTorch, LangChain, computer vision (YOLO, RF-DETR, ByteTrack), LLM tooling Agent infrastructure: MCP servers and protocol bridging, agent-skill scanning, static analysis AI evaluation: agent benchmarking, ground-truth case design, statistical scoring Backend: Django REST, FastAPI, PostgreSQL, Redis, Docker Web: React, Next.js, TypeScript, TailwindCSS, Vue IoT/Embedded: nRF54L15, BLE, Zigbee, Home Assistant, MQTT
mcp-uplift Protocol bridge that wraps legacy stdio MCP servers for the stateless MCP spec. MRTR (mid-request translation and resume) covers sampling, elicitation, and roots, which existing proxies drop.
ClawVet (npm) MCP and agent-skill security scanner. Static triage first, LLM adjudication second. Earlier versions have been used as a baseline in third-party agent-security benchmarks.
overruled (PyPI, GitHub Action) Verdict auditor for AI SOC agents. Replays ground-truth cases and grades the rulings. No LLM anywhere in the grading path.
Sift (npm) Semantic triage layer for AI-agent security scanner output. Takes SARIF, returns classified findings.
Ferry (npm) Model-migration CLI. Compares quality and cost delta across LLMs before you switch.
IoT home security system nRF54L15 with BLE and Zigbee mesh for fingerprint auth and sensors.
"Comparative Assessment of YOLO Nano Architectures for High-Speed Steel Detection", ASTRJ, 2026. DOI
Stats
Open to remote AI/full stack roles, IoT projects, and consulting.