Production GenAI systems
RAG pipelines, retrievers, rerankers, validation flows, and reusable LangChain components.
AI/ML Engineer / Backend Systems / Applied Research
Building production AI systems and advancing research.
I am Wassim Nijaoui, currently working remotely as an AI Engineer at XCoding Solutions. My recent experience spans production RAG systems, knowledge-graph and ingestion-based chatbots, Java microservices, anomaly detection, research engineering, and startup product work across XCoding Solutions, Ericsson, CGI, Ciena, Cisco, Concordia, and my own ventures.
Ericsson
XCoding Solutions
CGI
Cisco
Ciena
Concordia
Teluq
Right now
Production GenAI, semantic search, reranking, and reusable AI components.
Snapshot
XCoding Solutions now, two Ericsson chapters, one year at CGI, one year at Ciena, and startup work on the side.
Profile
Overview
My work sits at the intersection of production AI, backend engineering, applied research, and product development, with experience across telecom, academia, and startup environments.
Backend and platform work
Spring Boot microservices, REST APIs, Kafka systems, monitoring automation, and ML-enabled operations.
Research to implementation
From university research and published work to shipped prototypes and production-facing tools.
Experience
Experience
Professional, startup, and research work organized in separate tracks for easier navigation.
Ericsson
Ericsson
AI/ML Engineer
- Designed production-scale user-facing RAG-based GenAI systems and continued supporting them after rollout.
- Built agentic and multi-agent workflows for retrieval, reranking, validation, and response orchestration.
- Improved semantic search depth and relevance through embedding optimization and retriever tuning.
- Delivered custom reusable LangChain AI components that accelerated prototyping and development.
AI/ML Engineer
- Engineered an anomaly detection system to identify failing Kubernetes pods and microservices in real time.
- Improved preprocessing speed by 200 percent and reduced model training time by 50 percent for the Failure Prediction PoC.
- Managed an Agile-based project flow and implemented tests and visualizations.
- Reduced data dimensionality by 40 percent while contributing to the product-facing implementation.
RAG
LangChain
Kubernetes
Anomaly Detection
Full-Stack
Software Engineer
CGI
- Built Java 8 and Java 21 Spring Boot microservices for low-latency telecom systems.
- Maintained Kafka and REST services handling 20M plus daily requests under TMF standards.
- Automated KPI monitoring and reduced recurring team effort by 80 percent per day.
Java
Spring Boot
Kafka
Microservices
Software Engineer Intern
Cisco
- Included as part of the core company timeline between Ciena and the first Ericsson internship.
- Ready for project scope, stack, and outcome details in the next update.
AI/ML and Full-Stack Engineer Intern
Ciena
- Created a meeting call analytics PoC integrated with AWS.
- Built AI models that improved QA workflows by over 500 percent of time spent on calls.
- Built product-facing interfaces and workflow integrations around the analytics platform.
- Managed ingestion pipelines for thousands of calls at over 90 percent lower cost.
AWS
Call Analytics
Explainable AI
Full-Stack
IT Support Specialist
CGI
- Solved incidents and support requests at both level 1 and level 2.
- Worked remotely and onsite to troubleshoot user and infrastructure problems.
- Built automations that reduced repetitive support work and improved team efficiency.
Support
Automation
L1
L2
AI Engineer
XCoding Solutions
- Designed end-to-end chatbot systems using rule-based logic, knowledge-graph reasoning, and ingestion-driven retrieval pipelines.
- Built workflows around Qwen and Qwen VL models for text, multimodal understanding, and task-specific assistant behavior.
- Used LangChain to structure retrieval, orchestration, prompt pipelines, and production-ready chatbot flows.
- Built a proctored cheating-detection system and developed off-screen gaze detection models from scratch, with over 90% accuracy.
Qwen
Qwen VL
LangChain
LLMs
AlloGenie
- Built around matching skilled workers with users who needed help through video calls or arranged meetings on the platform.
- Represents an early product-focused startup effort tied to marketplace and service matching workflows.
InsightWearAI
- Led the startup and won more than 6 awards across competitions and innovation programs.
- Created an AI-powered watch and necklace designed to help blind users navigate their surroundings.
- Focused on affordable assistive technologies rather than premium inaccessible hardware.
Founder-Led
Assistive AI
Wearables
6+ Awards
Parklive
- Helped users find the nearest available parking spot and understand parking signs more clearly.
- My work focused mainly on route optimization and decision logic around parking discovery.
Routing
Optimization
Mobility
Research Collaboration with Desjardins, Occupancy Estimation
- Worked on occupancy estimation from thermal imagery, air quality signals, data engineering, and system design.
- Combined sensing, pipelines, and model work to estimate overall occupancy in practical settings.
Thermal Images
Data Engineering
System Design
Occupancy
NSERC Project, Anomaly Detection for Smart Homes
- Worked on anomaly detection in smart-home settings with data augmentation in the NILM domain.
- This work feeds directly into the ICONIP 2024 publication highlighted elsewhere on the site.
AI Researcher and Engineer, Air Quality Project
Concordia
- Handled hardware including PCB design, case design, circuits, wiring, and the physical build of air quality monitors.
- Designed system architecture for servers and AQ devices, including device-side code in C.
- Deployed, monitored, and maintained the overall system over long-running research use.
- Designed and deployed more than 20 high-resolution air-quality monitors that stayed operational for over 4 years.
Hardware
PCB Design
C
Deployment
NSERC Project, Object Counting and Detection
- Modified DETR and fine-tuned it for object counting and detection use cases.
- Focused on improving a research model for the actual task rather than only benchmarking it.
NSERC Project, Semantic Segmentation on Road Scenes
- Developed a multi-model approach for semantic segmentation on road scenes.
- Worked on experimentation and model design for perception tasks in driving-related imagery.
Education & Publications
Education & Publications
Degrees, publication work, and academic milestones that support the broader engineering profile.
Master of Applied Science
Artificial Intelligence
Concordia University, Montreal
- Timeline: 01/2025 - 05/2026.
- Focus: model compression for cloud and edge systems.
Bachelor of Engineering
Software Engineering
Concordia University, Montreal
- Timeline: 01/2020 - 12/2024.
- Award in Engineering Excellence.
ICONIP 2024
Behavior-Driven Data Augmentation for NILM, published with Springer Nature.
- Published during my undergraduate studies.
Awards and Recognition
Awards and Recognition
Selected awards, competition results, and publication recognition.
NSERC Undergraduate Student Research Award
Received 3 times: 2021, 2022, and 2024.
Ferique Engineering Competition
3 awards listed in the current resume.
Gina Cody Innovation Fund
1st place, 3 times: 2022, 2023, and 2024.
ICONIP 2024 Paper
Published work highlighted as part of both the academic and recognition profile.
Featured Projects
Solar Site Feasibility - Multi-Agent PoC
A recent GenAI proof of concept that combines multi-agent orchestration, public-source research, permitting workflows, and scoring logic into a practical Go or No-Go solar feasibility tool.
Address-level feasibility scoring with modular AI agents
Built as a Streamlit and LangGraph proof of concept, the system evaluates one or more addresses, runs independent agents for research and permitting, and can optionally layer in irradiance, economics, grid, risk, and web context before producing a final feasibility score and decision.
- Uses LangGraph state orchestration so agents run independently and fan in to a shared scoring stage.
- Generates human-readable justifications plus a professional permit checklist PDF.
- Supports both deterministic heuristic scoring and LLM-assisted scoring for the final Go or No-Go recommendation.
LangGraph
Streamlit
Multi-Agent AI
PDF Generation
Rotates automatically every few seconds. You can also switch projects manually.
Capabilities
Technical Focus
Core areas that best represent the work shown across the site.
GenAI and LLM systems
Agentic AI, multi-agent workflows, fine-tuning, semantic search depth, OpenAI APIs, LangChain, and Langflow.
Backend and APIs
FastAPI, Flask, Django, Spring Boot, REST APIs, Swagger, and event-driven service design.
Cloud and MLOps
AWS, Docker, Kubernetes, Jenkins, GitHub Actions, CI/CD automation, and production support workflows.
Contact
Email
Available for AI engineering, research collaboration, and product-focused work.
The site now reflects the main experience, research, and project work. If you want to connect about roles, collaborations, or technical projects, the best links are below.
Phone
Focus
LLMs, backend, applied AI