Wajahat Ahmad

Wajahat Ahmad, Software Engineer, Backend & AI, Brussels

Reliable backends. Human-approved AI.

Distributed messaging in .NET, Java/Spring Boot platforms, and AI agents that stay human-approved.

View projects Resume (PDF)

Backend engineer with a systems mindset.

I started in distributed systems, building messaging and state transfer inside NCache, a caching platform used in production worldwide. Today I also build LLM applications: agents, RAG, and automations that run in production.

Based in
Brussels, Belgium
Work status
Eligible to work in Belgium
Languages
English (professional), French (elementary, improving), Urdu (native)
Now
Software Developer, AI & Automation at Procyo

Reliability

Systems that keep working through failover, retries, and bad input.

clientnode 1node 2node 3

Three nodes share the load.

Security

Least privilege, validated input, and secrets kept out of code.

Nothing reaches the system until it passes the boundary check.

Simplicity

The smallest design that solves the problem, so the next engineer can change it.

Six parts, nine dependencies.

The model proposes. A person approves.

Every agent action maps to a step a human can check.

What I work with

Built Pub/Sub messaging inside NCache in C# / .NET, with client SDKs over Protocol Buffers and RPC.

Built the ERTP recruitment platform in Java and Spring Boot, tested with JUnit, now live in production.

Sole engineer on the Procyo CRM in Python and Odoo, deployed through GitHub Actions to Azure.

Agents with tool calling, structured outputs, and RAG in production at ERTP and Procyo, traced in Langfuse.

  1. Backend

    Used at ERTP, Procyo, NCache, ZAROORAT

    • C# / .NET Core
    • Java / Spring Boot
    • Python
    • REST APIs
    • Protocol Buffers + RPC
    • Odoo

  2. AI engineering

    Used at ERTP, Procyo

    • LLM agents
    • Tool calling
    • Structured outputs
    • RAG
    • Multi-agent workflows
    • Langfuse
    • Anthropic Claude API
    • n8n

  3. Data & messaging

    Used at ERTP, Procyo, NCache, EcoGuide

    • PostgreSQL
    • MongoDB
    • Redis
    • Distributed caching
    • Pub/Sub (Kafka patterns)
    • Vector databases

  4. Frontend

    Used at ERTP, NCache, ZAROORAT

    • Angular
    • TypeScript
    • Blazor
    • React Native

  5. Delivery

    Used at ERTP, Procyo, NCache

    • Docker
    • Azure
    • GitHub Actions
    • CI/CD
    • JUnit
    • Claude Code
    • Cursor

Where each skill was used

Hover or tap a skill or a project.

Backend

Data & messaging

Frontend

AI engineering

Delivery

Selected work

ERTP, MSc thesis with UCLouvain. Live in production, closed source.

AI Recruitment Automation Platform

An LLM platform that automates recruitment work: CV parsing, personality assessment, and evaluation of open-ended answers. Every agent action maps to a concrete step that a recruiter approves.

Java, Spring Boot, Angular, PostgreSQL, Anthropic Claude API, Langfuse, JUnit

ERTP system architecture: admin and candidate call a REST API, results are stored in a database, an async evaluation worker calls LLM providers through a model adapter layer.CV parsing flow: the candidate uploads a CV, text is extracted, an LLM extracts structured data into an admin-defined template, the candidate reviews and submits.Open-answer evaluation flow: the backend launches an async worker that prompts several models through a model adapter layer and writes results to the database for admin review.

Original diagrams, opens full size.

System architecture from the thesis: REST API, async evaluation worker, and a model adapter layer over several LLM providers.
ownslisted bytenantquoteOwnerProperty (estate.property)BrokerLease (estate.lease)TenantTicket (estate.ticket)SupplierInvoiceDepositInspectionApprovalSupplier Invoice
Core business entities in the Odoo CRM and how they relate.

Procyo, client work

AI-augmented Property CRM

Odoo CRM built from scratch as the sole engineer, with RAG and agent workflows in production. Legacy data migrated over XML-RPC. CI/CD deploys only validated changes to Azure.

Python, Odoo, PostgreSQL, Docker, Azure, GitHub Actions

NCache architecture. Diagram: Alachisoft.

Alachisoft

NCache Pub/Sub Messaging

Kafka-style topics, publishing, polling, and subscriptions inside a distributed cache cluster, consistent across partitioned, replicated, and mirrored topologies.

C#, .NET, Java, Protocol Buffers

UCLouvain

EcoGuide

Two-tier search: a keyword index for exact matches plus a vector-database layer for meaning-based queries.

.NET, MongoDB, vector database

ZAROORAT project poster with the explore and profile screens of the app.

Bachelor's final year project

ZAROORAT

Marketplace app with in-app messaging and ML services for category prediction, price suggestion, and Roman-Urdu profanity filtering.

React Native, Firebase, Python, FastAPI

Homepage of Clean Tech Utilities with the headline Spotless Spaces Start With Clean Tech.

Client website

Clean Tech Utilities

Marketing website for a commercial cleaning company working in Ireland and the UK, with offices in Dublin and London.

Next.js

Homepage of the Spade rebuild, a fintech marketing site.

Learning project

Spade site rebuild

A rebuild of a fintech marketing site, made to practise front-end layout and typography in Next.js.

Next.js

Experience and education

  1. Work, Mar 2026 - present, Belgium

    Procyo

    Software Developer, AI & Automation

    • Sole engineer on a custom property-management CRM in Python/Odoo for a real-estate client, from design to deployment.
    • RAG, LLM agent workflows, and automations running in production.
    • Migrated a legacy system to new Odoo models with XML-RPC scripts. CI/CD with GitHub Actions: the test branch rebuilds a test instance, and only validated changes reach production on Azure.

    Python, Odoo, PostgreSQL, Docker, Azure, GitHub Actions, XML-RPC

  2. Work, Aug 2025 - Jun 2026, Belgium

    ERTP

    AI Engineer (MSc thesis, UCLouvain industry collaboration)

    • Built an AI-driven recruitment automation platform on LLM APIs, now live in production.
    • Production LLM integration: model selection, token-cost tracking, prompt tracing with Langfuse, streaming responses.
    • Took CV parsing, personality assessment, and automated evaluation of open-ended answers from R&D prototype to tested full-stack production.

    Java, Spring Boot, Angular, PostgreSQL, Anthropic Claude API, Langfuse, JUnit, Maven

  3. Education, 2024 - 2026, Belgium

    UCLouvain

    MSc Computer Science

    Options: Software Engineering & Programming Systems; Networks & Systems. Thesis supervisor: Prof. Kim Mens.

  4. Work, Jul 2023 - Sep 2024, Pakistan

    Alachisoft

    Software Engineer

    • Designed and built a Kafka-style Pub/Sub messaging system inside the NCache cluster: topics, publishing, polling, subscriptions.
    • Fixed state transfer across partitioned, replicated, and mirrored topologies and failover, keeping messages and offsets consistent.
    • Client SDKs in .NET, Java, and Python over Protocol Buffers and RPC. As Chief Trainee Officer, ran weekly architecture training and reviewed code from junior and senior engineers.

    C#, .NET, Java, Python, Redis, Protocol Buffers, Docker

  5. Work, 2022

    RadicalX

    Product Development Intern

  6. Education, 2019 - 2023, Pakistan

    COMSATS University Islamabad

    BSc Computer Science

Let's build something that doesn't break in production.

Open to backend and AI engineering roles in Belgium, and to AI automation work for SMEs.

GitHubgithub.com/wajahatiLinkedInin/wajahati-ahmadMediumwajahati.medium.comResume (PDF)

Tell me about the role, team, or project.