Lamis Ghoualmi, PhD

About Me

I am a Data & AI solutions architect and applied AI engineer who turns complex business problems into scalable, governed, production-grade data and AI platforms. I design end-to-end architectures on Microsoft Fabric — ingestion, Lakehouse and Medallion layers, Spark, semantic models, and Power BI — and I embed AI directly into the data foundation, so that insights are automated, persistent, and trusted rather than locked inside a chat window.

I lead work from idea to roadmap to delivery: framing the problem with stakeholders, authoring target architectures and proofs of concept, setting naming standards and reusable templates, and acting as the technical escalation point when reporting, integration, or analytics issues get hard.

My foundation is research. I hold a PhD in Computer Science, awarded With High Honors, with work in bio-inspired AI, biometric authentication, and machine-learning optimization, published with IEEE, Elsevier, and Springer (164+ citations, h-index 7). That rigor shapes how I engineer: clear hypotheses, measurable validation, confidence scoring, and humans kept in the loop.

I am a frequent speaker at professional and research conferences, including EDUCAUSE, the Tennessee Higher Education IT Symposium (THEITS), the DATA xUTK Conference, and the UT Innovative Technologies Symposium, presenting on AI-driven data quality, embedded AI for decision intelligence, and agentic analytics.

I Architect

Modern, secure data & AI architectures aligned with business goals.

I Lead

Roadmaps, standards, reusable templates, and technical escalation.

I Engineer

Reliable pipelines, models, and analytics solutions that scale.

I Innovate with AI

LLMs, RAG, and agents applied to real problems, with governance built in.

I Modernize

Bridging legacy systems (Oracle, PL/SQL, ODI) to the modern cloud.

I Communicate

Turning architecture and data into clear narratives for decision-makers.

Skills & Technologies

Microsoft Fabric
Lakehouse & Spark
Semantic Models
Power BI
Microsoft Foundry
Azure Logic Apps
Fabric Apps
AI Web Apps
Data Factory
ETL Pipelines
Agentic AI
RAG & Embeddings
Azure OpenAI
Anthropic
LLMs
Machine Learning
Responsible AI
Python
SQL / Oracle
Oracle Data Integrator
Data Visualization
Biometrics

Experience

AI & Data Architect (Senior Software Analyst IV)

University of Tennessee, Knoxville  ·  Full-time

Jun 2023 – Present  ·  Knoxville, Tennessee · Hybrid

  • Led the end-to-end design and delivery of an AI-integrated project portfolio analytics platform on Microsoft Fabric — project management platform sources, Dataflows, OneLake Lakehouse, Spark, Direct Lake semantic models, and a role-secured Power BI app.
  • Conceived and built an embedded AI intelligence layer (Azure OpenAI, SynapseML, Fabric AI Functions): automated project summaries, risk classification, sentiment scoring, semantic similarity across projects, and agentic alerts to leaders through Teams and email.
  • Authored roadmaps and proofs of concept for reporting modernization, data operations observability, and AI-driven data enrichment on Microsoft Fabric.
  • Developed enterprise ETL with ODI, PL/SQL, SQL Server, and Fabric — data marts, views, stored procedures, jobs, and reporting datasets.
  • Built and deployed Power BI dashboards and apps for project management, enrollment management, HR, and operational reporting.
  • Serve as a technical escalation point for complex reporting, integration, and analytics issues, and partner with IT teams and senior stakeholders on requirements and rollout.

AI & ML Research Scientist (Biometrics & Healthcare Analytics)

University Abdelhamid MEHRI Constantine 2  ·  Contract

Jan 2021 – Jan 2023  ·  Remote

  • Engineered a multi-objective optimization framework for palm vein biometric authentication, jointly optimizing accuracy and feature-vector compactness through a custom AI-driven loss function.
  • Implemented bio-inspired metaheuristic optimizers (PSO, ABC) for hyperparameter tuning and feature subset selection.
  • Built an end-to-end data and deep learning pipeline for chronic kidney disease prediction, with feature engineering and model optimization for diagnostic accuracy.
  • Built automated ML pipelines for heart disease prognosis on the 2020 CDC dataset, tuning classification models for clinical decision support.

Postdoctoral Research Associate (AI-Driven Environmental Exposome & Cardiovascular Health Analytics)

University of Tennessee, Knoxville  ·  Contract

Nov 2019 – Dec 2020  ·  Knoxville, Tennessee · Hybrid

  • Led the data engineering and ML pipeline on large-scale public health exposome (PHE) data at the Langston Laboratory, studying environmental impacts on population health.
  • Built a robust preprocessing pipeline for duplicates, missing values, outliers, and low-variance features.
  • Designed a feature selection framework using Mutual Information and Decision Trees, validated with Pearson correlation analysis for multicollinearity.
  • Engineered a graph-theoretic paraclique representation giving a novel topological view of environmental feature interactions.

Assistant Professor of Computer Science

University Abdelhamid MEHRI Constantine 2  ·  Full-time

Sep 2017 – Nov 2019  ·  Constantine, Algeria · On-site

  • Designed and delivered undergraduate and graduate courses in Algorithms, Network Fundamentals, Formal Languages, and Multimedia Image Processing, making lectures engaging while meeting curriculum objectives.
  • Led graduate research supervision for an AI-based Vehicle License Plate Recognition system, giving technical direction on computer vision, machine learning, experimental methodology, and thesis development.
  • Supervised and mentored undergraduate and graduate students across the full software engineering and research lifecycle: project planning, system design, implementation, testing, documentation, and preparation for the dissertation defense.
  • Mentored undergraduate capstone projects, including a collaborative web platform for student knowledge sharing, activity management, and real-time chat.
  • Coached students in research methodology, critical thinking, problem solving, and software engineering best practices, encouraging innovation and independent learning.
  • Evaluated academic performance and advised students on research, technical challenges, and career development.
  • Collaborated with faculty to improve course content and maintain high academic standards across the Computer Science program.

Featured Solutions & Architecture

Solutions I have designed end to end, from the business problem to the target architecture, the key design decisions, and a working proof of concept or production delivery. Select View architecture on any card to see the design and the tools used.

01 AI Solution Initiative · Strategy & Architecture

AI-Powered Project Portfolio Intelligence

Business problem: Dashboards report the past, while early warning signs stay buried in status updates and risk logs. Vendor AI is locked inside individual tools, and chat copilots have to be prompted again before every meeting.

Solution: A production, end-to-end Microsoft Fabric architecture in which AI is part of the data model. Insights are generated on a schedule, stored as data, and served to leaders through a governed semantic model, a Power BI app, and agentic alerts.

Value: Earlier risk visibility, portfolio-wide consistency, and automated insight with no extra AI licensing.

  • Microsoft Fabric
  • Azure OpenAI
  • SynapseML
  • Data Agent
  • RAG
  • Agentic AI
02 AI Solution Initiative · Strategy & Architecture

AI-Driven Data Cleaning & Semantic Standardization

Business problem: Data cleaning can consume up to 80% of an analyst's time. Free-text notes and inconsistent categorical fields are the hardest part, and they affect reporting and governance.

Solution: A governed Medallion pipeline in Fabric. LLMs correct free text at scale, while RAG, embeddings, and cosine similarity map messy values to a curated reference library. Confidence scoring sends uncertain cases to people. It grew from my 2024 Python/Streamlit prototype into a scalable design.

Value: Less manual cleaning, faster data readiness, and more trust in analytics, with human expertise kept in the loop.

  • Medallion
  • Embeddings
  • Cosine Similarity
  • Human-in-the-loop
  • Python
  • Streamlit
03 AI Solution Initiative · Strategy & Architecture

Survey Feedback Intelligence

Business problem: Open-ended survey comments hold valuable insight, but manual analysis can take months, and chatbot answers change with every prompt.

Solution: A Fabric data app built on engineered prompts that run the same way on every comment and every term. AI classifies sentiment and themes, summarizes feedback by group and term, and a RAG-grounded executive advisor links findings to strategic goals.

Value: Reliable, comparable insight in place of months of manual review, with responsible AI built in: no invented numbers, privacy protected, every result verifiable.

  • Fabric Data App
  • Rayfin SDK
  • AI Functions
  • RAG
  • Responsible AI
04 AI Solution Initiative · Strategy & Architecture

AI Major Recommendation for Freshmen

Business problem: Choosing a major is one of the hardest early decisions for students, and traditional advising can feel impersonal.

Solution: An LLM-powered app that turns a student's interests, skills, and goals into personalized major, program, and career recommendations, with clear explanations. The roadmap adds a machine-learning model trained on survey data.

  Try the app
  • OpenAI
  • Python
  • Streamlit
  • Recommendation

More Solution Work

  • Legacy Reporting Modernization Roadmap & PoC
  • Cross-Platform Data Operations Observability Roadmap & PoC
  • AI Enrichment with User Data Functions Roadmap & PoC
  • And more: ongoing Data & AI initiatives In progress

Speaking & Presentations

I regularly present at professional and research conferences — national, state, and institutional — sharing original ideas, architectures, and live demos. Selected recent talks:

  • AI in the data foundation
  • LLM-driven data quality
  • Agentic AI & decision intelligence
  • Responsible AI & governance
  • From prototype to production

Enhancing Data Quality in Higher Education: AI-Driven Automated Data Cleaning

DATA xUTK Conference · University of Tennessee

2026  ·  Talk + 2 live demos

  • How a Python/Streamlit prototype grew into a scalable, governance-aware Medallion pipeline in Microsoft Fabric.
  • Live demos: AI cleaning of messy free-text comments, and semantic standardization of job titles with RAG, embeddings, cosine similarity, and human review for low-confidence matches.

From Status to Strategy: AI-Powered Project Management Insights with Microsoft Fabric & Power BI

UT Project Management Community of Practice

Spring 2026  ·  Talk + live demo

  • Moving project management from reporting to decision intelligence: AI summaries, risk insights, project similarity, and agentic alerts for emerging risks.

From Status to Strategy: AI-Powered Executive Project Insights with Microsoft Fabric & Power BI

UT Innovative Technologies Symposium

2026  ·  Talk + live demo

  • Why vendor AI and chat copilots fall short for enterprise reporting, and how embedding AI in the data foundation makes insights persistent, reusable, and governed.
  • Compared AI integrated into the data foundation with AI through a Fabric Data Agent.

Enhancing Data Quality in Higher Education: AI-Driven Automated Data Cleaning

EDUCAUSE Annual Conference

Oct 2025  ·  National conference

  • Presented an AI-assisted approach to cleaning and standardizing messy text with LLMs and fuzzy matching, and its practical value for data quality.

Discovering Your Path: Tailored Major Recommendations for Freshmen Using AI

UT Innovative Technologies Symposium

2025  ·  Talk + app demo

  • Designed and demonstrated an LLM-powered app that recommends majors, programs, and career paths from a student's interests and skills.

AI-Assisted Data Cleaning & Standardization (prototype)

UT IT Symposium

Jan 2024  ·  Talk + demo

  • First demonstration of the idea: a Python and Streamlit proof of concept using the OpenAI API and fuzzy text matching to clean and standardize data.

Research

PhD in Computer Science, awarded With High Honors. My research sits where AI, optimization, and data meet, and it still guides how I design systems today.

164+Citations
7h-index
3Publishers: IEEE · Elsevier · Springer

Biometric Authentication

Ear, iris, face, and palm vein recognition: SIFT and DWT feature extraction, multimodal feature-level fusion, and multi-objective optimization that balances accuracy against compact feature vectors.

Bio-Inspired AI & Metaheuristics

Artificial Bee Colony, discrete Particle Swarm Optimization (including opposition-based learning), and genetic algorithms for feature selection and optimization.

Healthcare Machine Learning

Prediction models for chronic kidney disease, heart disease, and diabetes, including decision-fusion approaches.

Environmental Exposome Analytics

Feature selection with Mutual Information and Decision Trees, plus graph-theoretic paraclique analysis of environmental factors in cardiovascular health.

Applied LLM Research

LLMs, RAG, and embeddings for data quality, semantic standardization, recommendation, and decision intelligence, with confidence scoring and human review.

Teaching & Mentoring

Former Assistant Professor of Computer Science: taught undergraduate and graduate courses, led graduate research supervision in computer vision and ML, and mentored capstone projects.

Selected Publications

  1. 2016

    An ear biometric system based on artificial bees and the scale invariant feature transform

    L. Ghoualmi, A. Draa, S. Chikhi · Expert Systems with Applications, 57, 49–61 · Journal

    68+ citations
  2. 2015

    An efficient feature selection scheme based on genetic algorithm for ear biometrics authentication

    L. Ghoualmi, A. Draa, S. Chikhi · 12th International Symposium on Programming and Systems (ISPS)

    27+ citations
  3. 2014

    A SIFT-based feature level fusion of iris and ear biometrics

    L. Ghoualmi, S. Chikhi, A. Draa · IAPR Workshop on Multimodal Pattern Recognition of Social Signals in Human-Computer Interaction

    22+ citations
  4. 2015

    Ear feature extraction using a DWT-SIFT hybrid

    L. Ghoualmi, A. Draa, S. Chikhi · Intelligent Data Analysis and Applications (Proceedings of the Second Euro-China Conference)

    14+ citations
  5. 2021

    A multi-objective optimization approach for palm vein feature selection based on the discrete PSO

    L. Ghoualmi, M.E.A. Benkechkache, A. Draa, S. Chikhi · IEEE 6th International Conference on Signal and Image Processing (ICSIP)

    8+ citations
  6. 2022

    Feature selection based on machine learning algorithms: a weighted score feature importance approach for facial authentication

    L. Ghoualmi, M.E.A. Benkechkache · 3rd International Informatics and Software Engineering Conference

    8+ citations
  7. 2022

    Machine learning models for automated heart disease prognosis

    L. Ghoualmi, M.E.A. Benkechkache, A.D. Ghoualmi · 3rd International Informatics and Software Engineering Conference

    7+ citations
  8. 2021

    An improved discrete particle swarm optimization based on opposition-based learning for palm vein authentication

    L. Ghoualmi, M.E.A. Benkechkache, A. Draa · International Conference on Electrical, Computer, Communications and Mechatronics Engineering

    5+ citations
  9. 2017 PhD thesis

Contact

I enjoy connecting with people who are interested in data & AI architecture, applied AI, and modern analytics on Microsoft Fabric, especially work that brings AI into the data platform. I welcome speaking invitations, research and professional collaborations, and conversations about these topics. The best way to reach me is LinkedIn.

Clicky