Open to opportunities

SM Jahid

@smjahid

Senior AI Systems Engineer specializing in LLM gateways and agent orchestration, with a proven track record of delivering high-impact AI solutions

Bangladesh

Message

What I'm looking for

I'm seeking a remote AI Systems Engineer / AI Infrastructure role focused on LLM platforms, MCP tooling, and agent orchestration — where I can bring 9 years of production engineering experience (3+ in AI) to own systems end-to-end, from architecture through deployment, in async remote-first teams.

AI Systems Engineer building production AI infrastructure across LLM platforms, MCP tooling, agent orchestration, edge AI, and AI SaaS.

Developed Serper-search-mcp v3 (Public), an MCP-native search server, resulting in ~1.2k monthly npm downloads. Enabled seamless tool discovery for any MCP agent without per-client integrations while adding 5 defense layers to protect agent context.
Developed Anthropic-proxy v4.6 (Public), a multi-provider LLM gateway, allowing non-Anthropic models (Gemini, DeepSeek, Groq) to replicate Claude’s coding and tool-calling behavior under 3s latency through intelligent adapter chains and fallback logic.
Built SMAGENT (Private, Pre-released), a policy-governed agent orchestration framework, solving unsafe agent actions. Implemented 15 hard safety blocks, FSM enforcement, and 200+ tests, ensuring every action passes policy checks before execution.
Engineered MedGemma FORENSIC (Public, Pre-released), an offline Android AI system running 4.3GB LLM+Vision+Audio models on a 4GB device. Reduced peak memory to under 417MB by identifying and fixing a critical JNI buffer leak, enabling stable operation.
SM Expense (Private, Live) multi-tenant SaaS featuring Offline Voice AI (audio never leaves device), 90% Offline OCR accuracy in under 5s, IRS Pub 970 tax workflows, and live payment infrastructure with Merchant of Record tax handling.

With 9 years of production engineering experience (including 3+ years in LLM infrastructure and agentic systems), I own end-to-end delivery from architecture to production deployment.

I’m available for overlap with US EST and UK GMT business hours when real-time collaboration is needed.

Live Links:

https://github.com/smjahid012/medgemma-forensic/blob/main/README.md

https://github.com/smjahid012/serper-search-mcp-server

https://github.com/smjahid012/anthropic-proxy

Experience

Work history, roles, and key accomplishments

Current

AI Engineer

Current

SMLabs AI

Jan 2023 - Present (3 years 6 months)

Lead AI engineer building production systems across offline Android inference, MCP tooling, multi-LLM gateways, and agent orchestration. Shipped an offline AI stack running 4.3GB of models in <417MB RAM, launched Serper Search MCP Results (~1.2k npm downloads/month), and delivered SM Expense SaaS with payments, OCR, and voice workflows while guiding architecture and implementation across projects.

MCP (Model Context Protocol)ONNX Runtime Docker Compose Anthropic: Claude Claude Agent SDK Gemini API Gemini CLI AI Integration LangGraph Multi Agent Orchestration WiFi Direct Nearby Connections Generative AI AI Engineering AI Tools OpenAI Next.js LLM Integrations Fine Tuning LLM Gateways

Freelance Developer & AI R&D

Independent Contractor

Dec 2021 - Dec 2023 (2 years)

Delivered 150+ client projects across real estate, e-commerce, and business systems with a 100% satisfaction across Upwork & Fiverr while transitioning into AI engineering. Built full-stack solutions that became foundations for early agentic AI systems. Led end-to-end delivery, mentored teams, and introduced reusable patterns + automation that clients reported back improved dev efficiency by ~25%.

Full-Stack Developer(In House)

Jibon Computer , 3g Telecom

Feb 2017 - Dec 2021 (4 years 10 months)

Grew from junior developer to senior full-stack engineer over 4+ years, delivering Laravel/PHP applications, payment integrations, REST APIs, and WordPress solutions for business and e-commerce clients. Established strong foundations in scalable backend architecture, client collaboration, and production delivery that directly support current work in building reliable, production-grade systems.