AI Automation Engineer
- Build agentic workflows and automation systems using FastAPI and GCP
- Design document-intelligence workflows processing 1,000+ docs/month
- Orchestrate Claude Opus and GPT-Codex workflows (12 min → 3 min runtime)
Mantaka Mahir
AI Transformation Expert
I help businesses design and implement agentic AI systems, automation infrastructure, and production workflows that reduce manual work and create measurable revenue impact.
- Role
- AI Transformation Expert
- Location
- Dhaka (Remote)
- Work Model
- Full-time / Contract
- Core Stack
- Python, FastAPI, Next.js, n8n, LangChain
About
A short introduction
Goal-driven AI Transformation Expert building RAG systems, AI agents, and production automation workflows. I focus on shipping measurable systems that reduce manual work, improve decision-making, and turn complex operations into scalable business outcomes.
Experience
Where I've worked
AI System Developer
- Build AI education systems for kids and women
- Create practical learning workflows that make AI tools easier to understand and use
AI Systems Engineer
- Built multimodal intake automation with Notion, n8n, and OpenAI
- Implemented manual-review routing, duplicate protection, and audit logging
- Passed acceptance tests for text, DOCX, PDF, Excel, image, and MP4/audio paths
Upwork — Freelance
Feb 2025 - Present
Freelance · Remote
AI Software Development
- Analyze existing systems to identify operational bottlenecks
- Implement targeted AI and automation solutions that directly increase revenue and efficiency
- Build agentic systems with Next.js, n8n, APIs, Python, and LLM tools
Momentum AI Consulting
Jan 2026 - Jun 2026
Contract · United States · Remote
AI System Developer
- Built Next.js interfaces and FastAPI/Django backend systems
- Delivered service pipelines, CRM workflows, and reporting systems
- Full-stack AI implementation with clean client handoff
AI Automation Engineer
- Built agentic content operations across research, publishing, translation, and media generation
- Connected Python, LangChain, n8n, FastAPI, Notion, OpenRouter, Claude, GPT, Gemini, and Kling
- Automated 14 daily workflows across eight platforms
Projects
Selected work
01
$250K/year content team replacement
Two AI agents running a full content operation for Openline: trend research, calendar planning, article and social drafting, image generation, multilingual publishing, video generation, and edit sync across eight platforms.
Openlinecontent infrastructure
14 daily workflows · 6 languages · 149 translations/run · 0 manual ops
Python · LangChain · n8n · FastAPI · Notion · OpenRouter · Claude · GPT · Gemini · Kling
02
Conover Industries multimodal intake automation
Production-ready intake automation that converts manual text, documents, spreadsheets, screenshots, and MP4/audio inputs from Notion into structured JSON, linked action items, review routing, duplicate protection, and audit logs.
multimodal intakeNotion operations
Multimodal intake · review routing · duplicate protection · audit logs
n8n · Notion API · OpenAI API · GPT-5 Mini · GPT-4.1 · GPT-4.1 Mini
03
Momentum AI Studio
Invite-only content production dashboard for podcast and video teams, connecting a Next.js app to n8n workflows for AI clip generation, project tracking, clip exports, social scheduling, thumbnails, descriptions, audio cleanup, and trim reports.
internal AI studiovideo workflows
5GB uploads · 14 languages · multi-platform posting · 15s processing refresh
Next.js 16 · TypeScript · React 19 · Tailwind CSS · NextAuth · n8n · Reap AI · CleanVoice AI · Google Drive · Google Sheets · Vercel
04
Multi-agent email attachment processing
Multi-agent pipeline for high-volume attachment processing with structured outputs and error-resistant data handling.
40x ROIhigh-volume processing
6-8 min → 8-12 sec · $23k-37k annual saved
Python · LangChain · n8n · API design
05
AI avatar video generation
Automated avatar video production from script to final output. Async polling, multi-API orchestration with HeyGen, Submagic, and OpenAI.
async pollingscript-to-video
2-4 hrs → 3-5 min · $31k-34k annual saved
Next.js · HeyGen · OpenAI · n8n
06
Content repurposing pipeline
Turns long-form videos into platform-specific social clips at scale. 100+ videos per month with 22x ROI.
100+ videos/month22x ROI
45-90 min → 4-8 min · $29k-32k annual saved
Python · AI APIs · n8n · Automation
07
Social distribution + lead capture
Automated scheduling, captions, comments, DMs, and lead capture with 5,000+ automated responses per month.
5,000+ responses/month20x ROI
15-30 min → 2 min/post · $23k-25k annual saved
n8n · ManyChat · API integrations · Next.js
08
Real Estate AI Video Engine
Converts one source video into eight platform-ready assets in approximately three minutes.
3 min runtimemulti-platform
1 source → 8 platform assets
n8n · AI APIs · Video processing
Skills
Tools and focus areas
Core
- Python
- C++
- Java
- JavaScript
- TypeScript
- FastAPI
- Next.js
- Django
- Node.js
- SQL
- Docker
- GCP
AI / ML
- LangChain
- LangGraph
- RAG
- Prompt engineering
- Embeddings
- Sentence-Transformers
- Claude
- GPT
- Gemini
- PyTorch
- Scikit-learn
- Hugging Face
- Pandas
- NumPy
Automation & Engineering
- n8n
- Make
- API design
- CI/CD
- GitHub
- Firebase
- Vercel
- MLOps
- System design
- Data structures & algorithms
Credentials
Certifications
01
n8n Verified Creator
Recognized for practical workflow and automation execution.
02
Stanford ML — Supervised Machine Learning
Regression and classification fundamentals.
03
AI Automation with Claude
Anthropic-certified Claude workflow design and deployment.
04
Claude Code in Action
Agentic coding with Claude Code for production systems.
05
MCP Advanced Topics
Model Context Protocol for agentic system design.
06
Google IT Automation
Python, Git/GitHub, and cloud operations training.
07
IBM AI Agents
AI agent fundamentals and orchestration.
Education
Bachelor of Engineering Computer Science
Military Institute of Science and Technology (MIST)
Coding CompetitionHackathonBusiness Case Studies
Now
What I'm focused on
Making AI systems that should exist but usually get stuck at the prototype stage. Simple is hard, measurable is harder, but both are worth it.
If you're deciding between a single-model workflow and a multi-agent system, that's usually where I do my best work.
Contact
Let's connect
Let's connect. I'm open to AI Transformation Expert roles, contract work, and collaboration on production AI systems.