Hi, I'm Shaan
AI Infrastructure Engineer
I build AI infrastructure: the agents, inference layers and pipelines that make LLMs actually useful.
From curious kid to AI infrastructure engineer
AWS-certified (Γ3) final-year Computer Science (AI) student and engineer focused on AI infrastructure and the full LLM lifecycle. I'm currently building AI-assistant infrastructure on placement at BlockAPT, and I run my own GPU / inference homelab. I communicate clearly, collaborate across distributed teams, and like owning hard problems end-to-end.
Quick facts
- Slough / London, UK
- English (Fluent) Β· Japanese (Conversational)
- AI Service team @ BlockAPT
Education
BSc Computer Science with Artificial Intelligence
2023 β 2027Queen Mary University of London
Steering modules toward AI, machine learning and security.
A-Levels in Mathematics, Physics, Computer Science
2021 β 2022ASCS School, Dubai
A*AA

High school graduation, Al Salam Community School, Dubai
The road so far
How a curious kid in the UAE became an AI infrastructure engineer.
2005 β 2021
2005 β 2021
Growing up in the UAE
Raised around the UAE's bet on technology, I fell for the idea that science and maths could reshape the world, and started pulling computers apart to understand how.
2021 β 2023
2021 β 2023
Silicon as a tool
Through A-Levels in Dubai I went deep on hardware, software, programming and early AI models. Computing stopped being a hobby and became how I wanted to make an impact.
2023 β present
2023 β present
Computer Science (AI) at Queen Mary
Moved to London to read BSc Computer Science with Artificial Intelligence at Queen Mary University of London, steering my modules toward AI, machine learning and security.
2025 β present
2025 β present
Building AI infra at BlockAPT
Joined the AI Service team at a London cybersecurity scale-up. I stopped working through tasks I was handed and started finding the problems worth solving.
Where I've shipped real work
My placement on the AI team showed me the kind of work I want to own.
BlockAPT
Data Science / AI Engineer (Industrial Placement)
London cybersecurity scale-up with one vendor-agnostic platform to Manage, Monitor, Automate and Respond, called by CISOs βthe system of all systems.β Founder & CTO: Marco Essomba.
Sep 2025 β Present
London, UK
AI Service team (BAP-AI)
- Engineered an MCP (Model Context Protocol) server giving the company's AI assistant a full tool layer over its cybersecurity platform: log ingestion, querying and automated response actions like blocking malicious IPs. Built as loosely-coupled microservices.
- Re-architected the assistant from a single-vendor integration into a vendor-agnostic inference layer, unlocking many model providers (including self-hosted models on local GPU) and removing vendor lock-in.
- Collaborated daily with the distributed BAP-AI team across time zones; my pull requests were reviewed and merged into the main codebase by senior engineers.
- Authored and delivered technical and proposal presentations to the team and CTO, explaining the motivation, design and phased timelines behind major architectural changes.
- Secured a contract with a major Japanese enterprise client (Canon) by stepping in to interpret in Japanese when the live-translation tool failed, turning a stalled discussion into a warm business relationship.
βSyed Shaan consistently exceeds expectations and demonstrates excellent integration within the team. Keep up the great work!β
Marco Essomba
Founder & CTO, BlockAPT
Initiative is currency
I stopped waiting to be handed problems and started going looking for them. That single shift earned the real trust and the real work.
Bring your whole self
A skill I'd filed away as Japanese, a skill I'd filed away as irrelevant to a βtech jobβ, became my most valuable contribution. The off-topic edges are often the point.
It's okay not to know everything
I joined not knowing microservices or MCP. Asking early and researching fast beat hiding the gaps every single time.
IT Department Intern Β· Masters Group
Supported the IT department across hardware, software and networking, troubleshooting issues and delivering practical solutions to business and client needs.
The stack I build with
A working toolkit across the LLM lifecycle, from model serving and agents to the cloud and backend that hold it all up.
AI / ML
Cloud & Infrastructure
Backend & Data
Languages & Tools
Things I've built
Agents, inference infrastructure, semantic search and a hackathon-winning password manager: a sample of what I ship.
IRCX
IRCv3 platform adapter for the Hermes LLM agent
Open-source adapter connecting Nous Research's Hermes agent to IRC: a full IRCv3 stack with SASL (PLAIN / EXTERNAL / SCRAM-256/512), an autonomous observe-mode and runtime tool-driven channel agency. 91 network-free tests run in GitHub Actions CI; live-validated on Libera.Chat and Rizon.
Vindex
Multimedia indexing & summarisation platform
Semantic search across video and audio: OpenAI Whisper transcription and Gemini 2.5 Flash summarisation indexed in Elasticsearch. GPU-accelerated, containerised and architected to scale on cloud GPU.
ZER0x
Zero-knowledge encrypted password manager
A fully client-side encrypted, zero-knowledge password manager built on the Internet Computer with Phantom wallet identity. Won 1st place (ICP track) and 2nd (Solana / Superteam UK) at Brunel Hack 2.5.
GPU & Inference Homelab
Self-hosted models on bare-metal GPU
Runs custom and pre-trained models on an RTX 4070 Ti, exposed from a home-GPU node over a WireGuard tunnel and reverse-proxied (Caddy / Nginx) through a public VPS. The backbone for my LLM automation pipelines and agentic tooling (OpenClaw, Hermes).
AI Music Recommender
Retrieval-and-ranking system (team lead)
Led a university team building a TensorFlow Recommenders retrieval-and-ranking system; ran weekly delivery, documentation and task delegation, and shipped despite mid-project changes to the Spotify API terms.
gemini-subagent
Delegate work to Gemini CLI from Claude Code
A Claude Code skill for delegating work to the Gemini CLI in headless mode: single-shot delegation, parallel co-development and real-time ACP steering, to save context and quota on long-running builds.
Proof of work
Triple AWS-certified, a blockchain hackathon win, and a few more along the way.



1st place, Brunel Hack 2.5 (ICP)
Brunel University of London Β· also 2nd, Solana/Superteam UK
2025AWS Certified Solutions Architect β Associate
Amazon Web Services
2025AWS Certified AI Practitioner
Amazon Web Services
2025AWS Certified Cloud Practitioner
Amazon Web Services
2025BCG GenAI Job Simulation
Forage
2025IEUK: Technology & Engineering
Bright Network
2025First runner-up, University Hackathon
Curtin University, Dubai
2023Let's build something together
I'm open to graduate and internship roles in AI / infrastructure, and always up for an interesting problem. The fastest way to reach me is email.
