recruiter view · no 3D, no game

Zain Raisan

AI Engineer · Abu Dhabi, UAE · UAE Golden Visa · 30 days notice

Projects

• Argus

  Computer vision · open source

  Thirty two camera streams, one laptop GPU. A C++20 detection pipeline. FFmpeg decode, TensorRT YOLO, overlay and NVENC encode, each in its own queue stage, keeping the video on the GPU from the first byte to the last.

  • 32 concurrent streams of YOLO11s on one 8 GB RTX 3070 Laptop, as demonstrated
  • 5 input and output protocols, from RTSP and HLS to MP4 and Matroska

  C++20 · TensorRT · FFmpeg · CMake · 32 streams · 1 GPU · 8 GB · decode to encode on device

  GitHub ↗

• Orb

  Android · Google Play

  A vision model in your pocket, in airplane mode. An offline assistant running quantized vision language models on the phone. It picks the Snapdragon NPU, GPU or CPU at runtime, transcribes with Whisper on device, and sends nothing anywhere. The inference layer underneath is open source as native-onnx.

  • 4 quantized VLMs, Qwen3-VL 2B and 4B, MiniCPM-V 4.6 and Gemma 4 E2B
  • 5 execution backends in native-onnx, from CPU to the Hexagon NPU

  React Native · ONNX Runtime · Qualcomm QNN · Android · Kotlin · NPU · GPU · CPU picked at runtime · 0 network calls

  Google Play ↗native-onnx ↗

• Deputy

  Agents · open source

  Any web page, as a typed API for agents. A Chrome extension and MCP server that grafts the W3C WebMCP standard onto sites that never adopted it. Four attributes on the page's own form, and Chromium generates the schema and runs the submission. Forms that change state are filled and left for a human to submit.

  • 1.63× fewer tokens than Claude driving Playwright MCP on the same form, in my benchmark
  • 115 tokens per page observation, against 14,170 for an accessibility snapshot

  TypeScript · Bun · MCP · Chrome · 216,888 → 133,366 tokens · 26.1 s → 17.2 s wall clock

  GitHub ↗Project page ↗

• globalmm

  Python · PyPI

  Teaching a local model to see, without training it. Generates a llama.cpp compatible vision projector for any local LLM. Instead of hours of gradient descent on caption data, it solves for one matrix in closed form, mapping frozen SigLIP patches into the model's own embedding space.

  • 1 least squares solve, a 1152 by d matrix, in seconds
  • 81 soft tokens per image in the target model's embedding space

  Python · NumPy · Hugging Face · 1 closed form solve · 1152 × d · seconds, not GPU hours

  GitHub ↗PyPI ↗

• Wirecopy

  Product · live

  A daily briefing you can actually finish. AI releases, security events and research for developers, built and run end to end. One Go binary as server or worker, Postgres with pgvector, and a Cloudflare edge read model, published every morning on GitHub Actions inside the free tier.

  • 2.5k+ feeds across 20 source families, licence checked per source in the pipeline
  • 0 cookies used to count reads. Nonce only CSP and SSRF refusal in the fetcher

  Go · TypeScript · Postgres · Redis · Cloudflare Workers · GitHub Actions · ~2,500 feeds · one edition a day · inside the free CI tier

  wirecopy.ai ↗

• Sourcing Agent

  Agents · private build

  Eight workflows that are not allowed to sign anything. A governed procurement platform I built alone. Agents draft, compare and route, while approvals run as durable human in the loop workflows on Temporal. They cannot approve spend, reject a supplier or sign a contract on their own, and every call goes through a gateway the customer controls.

  • 376 backend tests, plus a 45 case offline evaluation suite in Arabic and English
  • 650 MiB of container memory saved by sharing model weights across workers

  FastAPI · PydanticAI · Temporal · Qdrant · Docker · 376 tests · 45 evals · 650 MiB of RAM given back

Production work, Abu Dhabi government

• Bilingual legal retrieval · < 2 s

  600+ Arabic and English legal documents, one vector store per language, self hosted Qwen embedding and reranker models on NVIDIA Triton.

• Agent analytics platform · 13

  Thirteen specialised agents over two government databases, serving visual analytics to 10,000+ internal users.

• City wide road defects · 1,000+

  A quantized Qwen3 VLM with an LLM verification layer, reading dashcam and CCTV footage across the city.

• Live video analytics · 1,000+

  YOLOv11l with TensorRT on CCTV and RTSP sources, republishing annotated streams.

• Demand onboarding platform · Lead · lead developer

  A conversational planner that elicits requirements, finds reusable work in existing contracts, and estimates budget and timeline before anything is built.

• API catalogue standard · MCP · initiated and led

  One OpenAPI standard across teams behind an automated lint gate, with conforming APIs exposable to agents as MCP servers.

• Model serving platform · 20+

  LLMs on vLLM, embeddings on Triton, 20+ containerised AI apps on AKS, observed with Grafana, Prometheus, Loki and Tempo.

• AI Champions platform · 300+ · primary developer

  Courses, admin workflows and chat over governed model endpoints for 300+ government employees.

More builds

• native-onnx · React Native bindings for ONNX Runtime with NPU backends
• zdc · An image codec with zero dependencies, DEFLATE and Huffman written from the spec
• Tango (alpha) · A minimal agent framework for Go
• AutoFlow · Generates GitHub Actions and GitLab CI from a TUI
• react-native-client · Resumable background downloads, streamed to disk natively
• Haykal (prototype) · Real time 3D avatar voice, first mouth movement under a second
• Daqa (pre launch) · Describe the data you need, get a traceable Data Pack
• MIGraphX Field Guide · A practical guide to AMD's inference engine
• egoshell · A Unix shell written in Go
• KeyStrike · A real time multiplayer typing race
• Exchartz · Skia charts for React Native
• tamduae.com · Marketing site for a Dubai real estate office

Experience and education

• AI Engineer

  Department of Municipality and Transport, Abu Dhabi · 2025 to now

  Production RAG, agents and computer vision. Lead developer of the demand onboarding platform.

• Technology Consultant Intern

  Deloitte, Dubai · 2024

  Clustered regulations with a T5 model and built a Three.js network topology viewer.

• BS Computer Science and Computer Engineering

  American University of Sharjah · 2024

  GPA 3.81. Published in IEEE Access.

• Software Engineering Intern

  Mohammed Bin Rashid University of Medicine, Dubai · 2023

  Automated a document pipeline, cutting manual processing time by 80%.

• Embedded Systems Intern

  Remal IoT, Sharjah · 2023

  Designed a PCB shield and ran an ESP32 workshop for 30+ students.

Publication. Dental Radiograph Analysis for Improved Diagnosis, IEEE Access, 2024