Skip to content

Akele Belay

Applied AI & Cloud Engineer

I build complete AI and data systems, from the data pipeline and the model to the API and the cloud it runs on.

5+ years across machine learning, cloud and DevOps. These days that mostly means LLM agents and RAG over operational data, digital twins fed by sensor streams, and water, agriculture and climate analytics at Nedamco Africa.

View resumeSelected work

Open to projects and roles · any time zone

Email meGitHubLinkedIn

Selected work

Systems I've built

Client and employer projects across AI, data, and cloud. Source code is private.

Trading Automation Dashboard screenshot
US-based client · Freelance

Trading Automation Dashboard

Trading platform with real-time market data from Databento, OHLCV aggregation, automated strategy execution across multiple brokers, and Stripe subscription billing.

System architecture
  1. Databento feed
  2. OHLCV aggregation
  3. Strategy engine
  4. Brokers
Next.jsTypeScriptWebSocketsAWS LambdaStripe
Nedamco Africa

Digital Twin Operations Agent

Sensor and SCADA data flows through Event Hubs into Azure Data Explorer and Fabric and keeps an Azure Digital Twins graph current. An MCP server exposes KQL queries and twin lookups as tools for an LLM agent.

System architecture
  1. Sensors
  2. Event Hubs
  3. ADX / KQL
  4. Digital Twins
  5. MCP agent
Azure Digital TwinsEvent HubsKQLMCPAzure OpenAI
US-based client · Freelance

Social Media Automation Agent

AI agent that manages social media pages: content generation, image optimization and handling incoming questions.

System architecture
  1. Brief
  2. LLM content
  3. Image optimization
  4. Ayrshare publish
PythonFastAPILangChainOpenAIAyrshare
US-based client · Freelance

AI Customer Agent

Multi-channel customer service agent that handles phone calls, emails and web chat. Calls come in through Amazon Connect and an Amazon Lex bot, emails through Amazon SES, and a Lambda agent answers using Amazon Bedrock.

System architecture
  1. Connect / SES / chat
  2. Amazon Lex
  3. Lambda agent
  4. Amazon Bedrock
  5. Reply
PythonAWS LambdaAmazon BedrockAmazon LexAmazon ConnectAmazon SESLangChain
MMCY Tech

Top Candidate Picker

API that embeds resumes into Qdrant and uses semantic search with an LLM to pick the best-matched candidates for a job description.

System architecture
  1. Resumes
  2. Embeddings
  3. Qdrant
  4. LLM ranking
PythonAWS LambdaQdrantLangChainOpenAI

Experience

Where I've worked

Employment, consulting and freelance work across AI, data, cloud and web.

Companies

Nedamco AfricaMMCY TechAdicol

  1. Nedamco Africa

    2022 – Present · Addis Ababa, Ethiopia

    • AI Engineer

      Jul 2025 – Present

      Building AI and data systems for water infrastructure, agriculture and climate. Digital twins on Azure fed by sensor data through Event Hubs into ADX and Fabric, LLM agents with MCP on top of operational data, and the cloud infrastructure and CI/CD behind them.

      • Connected 10+ SCADA and water-quality sensors across 2 water sites, Dire Dawa and Harar, to a live Azure Digital Twins graph
      • Gave operators plain-language access to live site data through an LLM agent with MCP tools over KQL and the twin

    • ML Application Developer

      Aug 2023 – Jun 2025

      On assignment at MMCY Tech

      Developed and deployed machine learning applications using Qdrant, GPT, LangChain, Amazon Textract and AWS Lambda. Led projects end to end, from data preprocessing to model integration and API deployment.

    • Cloud Consultant

      2022 – Jun 2025

      Ran cloud readiness assessments and designed cloud architectures aligned with client goals, with roadmaps for secure, cost-effective adoption.

  2. iWork Technologies

    Feb 2024 – Apr 2024 · USA · Remote

    • DevOps Engineer

      Built and maintained scalable APIs on AWS using EC2, Application Load Balancer, RDS and Auto Scaling across availability zones, with CI/CD pipelines for automated deployments.

  3. Upwork

    Jul 2024 – Present · Remote

    • Freelance AI Engineer

      Python and FastAPI backends for clients in the US and elsewhere: multi-agent systems, autonomous AI agents and trading bots, designed, built and deployed end to end. That includes ML pipelines, real-time market data streamed over WebSockets through a Centrifugo broker, and the infrastructure on AWS and GCP.

What I work on

AI that works with real-world systems

My work sits across a few connected areas. A single project often touches most of them, from sensor data or documents to a deployed, monitored application.

AI & LLM engineering

RAG systems, LLM agents and MCP servers on OpenAI, Azure OpenAI and Azure AI Foundry. Embeddings and semantic search, plus document, resume and translation pipelines.

RAGAgentsMCPLangChainLangGraphQdrant

Data & real-time systems

Event-driven and streaming pipelines, from sensor events to market data. Designed so the same system can keep up as the number of users grows.

Event HubsKQLRedisWebSocketsDatabento

Cloud & DevOps

Mostly Azure, also AWS. I usually design the infrastructure and deployment pipeline along with the application, and keep an eye on cost.

AzureAWSTerraformKubernetesDockerCI/CD

Digital twins & IoT

Azure Digital Twins fed by SCADA and sensor data, with analytics and AI agents built on top of the twin's event streams.

Azure Digital TwinsSCADAADXFabric

Climate, water & agriculture

At Nedamco Africa: turning environmental, operational and satellite data into information people can act on.

WaterAgriculturePlanetary Computer
Exploring

Geospatial & GIS

Working out where GIS fits inside AI and data platforms, evaluating ArcGIS Location Platform, the ArcGIS APIs and Mapbox.

ArcGISMapboxGeospatial data
Backend & full-stack
PythonFastAPIFlaskNode.jsTypeScriptNestJSReactNext.jsPostgreSQLDocker

01

End-to-end ownership

I often design the application, the data flow, the infrastructure and the deployment pipeline together, so the pieces fit in production.

02

Practical architecture

Managed services where they make sense, and designs sized to the real load and budget rather than the biggest possible one.

03

Clear communication

Requirements agreed up front, regular updates across time zones, and documentation and handover so your team can run what we built.

Credentials

Education and certifications

BSc in Software Engineering

Addis Ababa Science and Technology University · 2017–2021

Software design, testing and project management.

Azure AI Apps and Agents Developer Associate

Active

Microsoft · 2026

Building AI apps and agents on Azure AI Foundry, Azure OpenAI and Azure AI services.

Verify credential

Azure DevOps Engineer Expert

Active

Microsoft · 2022

People, processes and tooling for continuous delivery on Azure.

Verify credential

AWS Certified Developer – Associate

Expired

Amazon Web Services · 2022

Developing, testing, deploying and debugging AWS cloud applications.

Verify credential

AWS Certified Solutions Architect – Associate

Expired

Amazon Web Services · 2022

Designing systems across the breadth of AWS services.

Verify credential

AI Programming with Python

Udacity Nanodegree · 2024

Python, NumPy, Pandas, Matplotlib, and building neural networks with PyTorch.

Testimonials

What clients say

From projects delivered through Upwork.

Great DevOps engineer and developer! Akele's work is thorough. He's diligent and communicates well. A great person to work with. Will hire again.
Bilel BouraouiUpwork client
Akele delivered great work on this project. He has great technical skills and has very wide and deep knowledge in many technical areas. He is very easy to work with and we would like to work with him again in the future.
Sean GladstoneUpwork client

Contact

Have a project or a role in mind?

I'm open to contract work and full-time roles, and happy to work across time zones. Tell me a bit about what you're building.

Send me a message