APPLIED_AI_ENGINEER

AI That Survives
Contact With Reality.

I'm Shamir Shakher, an applied AI engineer who builds assistants, vision systems, and data pipelines that plug into real business operations.

I care less about impressive demos and more about systems that still work on a Tuesday afternoon when the data is messy.

What I Do.

I split my time between Bangladesh and China, engineering depth on one side, industrial deployment on the other. The through line is the same: AI that reads from live systems, not static PDFs.

Commerce & Trade

Sales & Support AI

Catalog RAG · Live Quotes · Human Handoff

Conversational agents tied to inventory and pricing rules. I design flows where automated replies, formal quotes, and human escalation share one source of truth, so sales teams stop reconciling spreadsheets after every chat.

Industry & Infrastructure

Operations AI

Production · Vision · Energy

Shift summaries from machine logs. Trainable cameras for whatever you need to detect. Hierarchical models when intake and output must balance across dozens of sensors. I build the glue between physical operations and readable intelligence.

Skills & Stack.

A toolkit shaped by shipping AI in production, not by collecting certificates.

Language & Reasoning

LLM Fine-Tuning RAG Pipelines Agent Orchestration Prompt Systems

Vision & Perception

Object Detection Custom Classifiers Edge Deployment RTSP / Video Pipelines

Data & Integration

Time-Series DBs IoT Telemetry Legacy API Wrappers ETL / Stream Processing

Engineering

Full-Stack Web Python / PHP Docker / CI-CD Full-Stack App

Domains

B2B Export / Wholesale Manufacturing Utilities & Energy Logistics & Warehousing

How I Measure Success

  • Deployed to production, not stuck in pilot
  • Grounded in live data with clear fallbacks
  • Operators can explain what it did and why

Selected Work.

Production AI platforms I've designed and shipped, each built to connect to real operations from day one.

Models Are Commodity.
Context Is Scarcity.

Calling an API is easy. Wiring AI into the systems that already run a business, ERP rows, RTSP feeds, meter telemetry, the quote sheet sales actually uses, that is the hard part.

I optimize for fewer handoffs, fewer hallucinations, and fewer projects that never leave the sandbox. Less theater. More Tuesday afternoon reliability.

That is the bar I hold myself to on every build.

Let's Talk.

Open to conversations about applied AI, production deployments, computer vision, and industrial data, whether you're exploring a build or want a second opinion on an architecture.

I'm Especially Interested In

  • • Teams connecting AI to live operational data
  • • Manufacturing, trade, utilities, and logistics use cases
  • • Vision or agent systems that need to survive production