Pixemantic, Tunis (2026)

Name: Pixemantic
Address: Tunis, 1053, TN
Telephone: +21695610011

29/06/2026

🚨 Last Chance to Join Pixemantic AI Lab – Summer Internship 2026! 🚀
Applications are closing soon!
If you're passionate about Artificial Intelligence, Machine Learning, Computer Vision, LLMs, Agentic AI, Healthcare AI, Robotics, OCR, Speech AI, Industrial AI, Digital Twins, or MLOps, this is your opportunity to work on real products with real clients.

At Pixemantic AI Lab, interns don't just learn—they contribute to innovative AI platforms used in healthcare, finance, insurance, manufacturing, energy, and smart cities.

🔬 This summer, you'll have the opportunity to work on projects including:

🤖 Agentic AI & Multi-Agent Systems
📄 PixOCR – Intelligent Document AI
🎙 PixVoice – Medical Speech-to-Text
❤️ PixHeart – AI for Cardiology
📊 PiXTable – Financial Document Intelligence
🏭 Industrial AI & Predictive Maintenance
🚁 Drone AI & Digital Twins
🌡️ Thermal & Hyperspectral Imaging
⚡ Smart Energy & Infrastructure AI
💬 RAG, LLMs & AI Assistants

You'll gain hands-on experience with Python, FastAPI, React, PyTorch, TensorFlow, NVIDIA GPUs, LangChain, LangGraph, LlamaIndex, YOLOv8, PaddleOCR, Docker, Kubernetes, MLflow, and many other state-of-the-art AI technologies.

📍 Location
Pixemantic AI Lab
Bloc C, A-29, Lac Malären
Tunis 1053, Tunisia

📅 Internship Period
1 July – 10 August 2026

📖 Discover our Summer Internship Projects 2026:
https://canva.link/r822fclqe1x2bm0

📩 Apply now
📧 [email protected]
📧 [email protected]
📞 +216 95 610 011

👉 Follow our LinkedIn page for updates:
https://www.linkedin.com/company/pixemantic/

I look forward to welcoming motivated students who are eager to innovate, publish research, and develop AI solutions with real-world impact.

Don't miss this opportunity—this is your last chance to join the Pixemantic AI Lab Summer Internship 2026! 🚀

05/06/2026

🚀 Another Scientific Milestone for Pixemantic AI Lab
We are proud to announce the publication of our latest research paper at the 2026 IEEE 5th International Conference on Computing and Machine Intelligence (ICMI 2026).
📄 High Precision Table Structure Recognition Using Transformer-Based Detection: Fine-Tuning TATR on Large-Scale Cropped Dataset
👨‍🔬 Authors:
Dr. Dr. Mohamed TOUATI , Mohamed Wiem Mkaouer, Laurent Nana, and Frikha Sultan.
This research represents months of dedicated work by the Pixemantic AI Lab team to advance the field of Intelligent Document Processing (IDP), OCR, and Document AI.
🔍 The Challenge
Organizations across banking, insurance, healthcare, government, legal services, and finance manage millions of documents containing complex tables. Extracting and understanding these tables accurately remains one of the most difficult challenges in document intelligence due to merged cells, irregular layouts, missing borders, and multi-level headers.
💡 Our Contribution
We developed a high-performance Table Structure Recognition (TSR) framework based on a fine-tuned Table Transformer (TATR) architecture trained on a massive dataset of more than 835,000 annotated table images.
The solution combines:
✅ Transformer-based detection
✅ Deep visual understanding
✅ Advanced table structure reconstruction
✅ Large-scale supervised learning
✅ High-precision row, column, header, and cell-span recognition
📊 Results
Our model achieved outstanding performance:
🏆 AP@[IoU=0.50:0.95] = 97.1%
🏆 [email protected] = 99.8%
🏆 [email protected] = 99.5%
These results demonstrate near state-of-the-art performance and confirm the effectiveness of transformer-based approaches for large-scale document understanding.
🌍 Real-World Impact
This research directly supports Pixemantic's vision of building next-generation AI solutions for:
• Intelligent Document Processing (IDP)
• OCR and Data Extraction
• Banking and Financial Automation
• Insurance Claims Processing
• Government Digitalization
• Enterprise Knowledge Management
• AI-Powered Document Understanding
At Pixemantic, we believe that research should not remain confined to laboratories. Our mission is to transform cutting-edge AI research into practical solutions that create measurable business value for organizations across Africa, the Middle East, Europe, and beyond.
👏 I would like to sincerely thank my co-authors, collaborators, researchers, engineers, interns, and the entire Pixemantic team for their dedication, expertise, and commitment to innovation.
DOI: 10.1109/ICMI68585.2026.11539825
hashtag hashtag hashtag hashtag

02/02/2026

We're Hiring for 2026! Pixemantic Internships & Final Year Projects
Full-Stack AI Intern | AI Software Engineer Intern | Software Engineer Intern | Cloud & Data Engineer Intern
Pixemantic – Tunis-based deep-tech startup leading in trustworthy AI (OCR, GenAI, Computer Vision, NLP) – is recruiting passionate students for 2026 internships and Final Year Projects (PFE)!
we're building industrial AI solutions for healthcare, finance, insurance, and smart construction. Join us to work on real products like PiXTable, PixVOICE, PixInsure, PixNFS, PixVerify, BIMEX, and PixOCR.
Featured Project (Top Priority Recruitment):
AI Extraction of Financial Tables & Balance Sheets (inspired by PiXTable)
📌 Description: Build an intelligent AI tool that automatically extracts, structures, and analyzes complex financial statements and tables from PDFs/scans. Includes data normalization, inconsistency detection, ERP-ready exports (Excel/JSON), and financial insights powered by LLMs. Direct impact on fintech, banking, and audit sectors in Tunisia & MENA.
⚙️ Tech Stack: OCR (EasyOCR or PixOCR), FinBERT, Machine Learning, PyTorch, Hugging Face, REST API (FastAPI), Export to Excel/JSON, LLM Finance
🏷️ Tags: hashtag hashtag hashtag hashtag hashtag
Other exciting projects available (details on request):
GenAI Medical Speech-to-Text & Clinical Documentation (PixVOICE)
Medical OCR for Hematology Reports (PixNFS)
Agentic AI Platform for BIM Analysis & Decision Support (BIMEX)
Vehicle Damage Detection from Images (PixInsure)
Profiles We're Looking For (4-6 month internships / PFE):
Full-Stack AI Intern: Frontend (React/Flask) + AI Backend + APIs
AI Software Engineer Intern: PyTorch/Hugging Face, GenAI models, Agentic workflows
Software Engineer Intern: Python/FastAPI, microservices, deployment
Cloud & Data Engineer Intern: Data pipelines, cloud platforms (AWS/GCP/Azure), MLOps, secure processing
Apply now!
Send your CV + motivation letter (English or French) to [email protected]
Subject: "Hiring 2026 – [Profile: Full-Stack AI / AI Software Engineer / Software Engineer / Cloud & Data Engineer] – [Preferred Project: PiXTable or other]"
Explore our solutions: https://pixemantic.com/solutions
Check out our Final Year Project Presentation for more inspiration:
https://www.canva.com/design/DAHAFaUNcTY/Ay-F-FvXTN_EV0ZKpvc5iA/view?utm_content=DAHAFaUNcTY&utm_campaign=designshare&utm_medium=link2&utm_source=uniquelinks&utlId=he375e6eca9
Let's build the future of trustworthy AI together! 💡🇹🇳
TunisiaAI

18/01/2026

PixInsure – Professional Automotive Audit Powered by AI 🚗🤖

🚀 Welcome to the future of automotive insurance and vehicle inspection.

In this demo, we present PixInsure, an advanced AI-powered automotive audit platform developed by Pixemantic, designed to transform how vehicle damage assessment, fraud detection, and repair estimation are performed.

🔍 What does PixInsure do?

PixInsure analyzes high-resolution vehicle images to deliver instant, reliable, and explainable results, even in real-world conditions.

⚙️ Key Capabilities Demonstrated in This Video:

✅ Authenticity Verification & Fraud Detection
Detects AI-generated or manipulated images using forensic analysis (ELA, texture analysis, artifact detection, metadata inspection).

✅ Vehicle Damage Detection
Automatically identifies dents, bumper damage, headlights, and body panel issues with severity scoring.

✅ Grounded Repair Cost Estimation
Provides realistic cost estimates based on damage type, severity, and market data.

✅ AI-Powered Digital Restoration
Generates a visual simulation of the vehicle’s condition after repair and maintenance.

✅ Explainable AI Reports
Clear technical reports designed for insurers, auditors, and automotive experts.

🏦 Who is PixInsure for?

Insurance companies

Automotive assessors & experts

Fleet & leasing companies

Mobility and InsurTech platforms

🌍 Why PixInsure?

In an era of digital claims and AI-generated content, trust is critical. PixInsure turns images into verifiable, auditable, and trustworthy evidence, reducing fraud and accelerating claim processing.

🔗 API-ready | Scalable | Built for real-world data
🧠 Developed & owned by Pixemantic

📩 Interested in a demo, partnership, or integration?
Contact us: [email protected]

31/12/2025

PiXTable : AI-Powered Balance Sheet & Financial Table Extraction by PixOCR | Built by Pixemantic
“Financial documents are full of valuable data but most of it is locked inside PDFs, scanned reports, and complex tables.”
PiXTable is designed to automatically extract balance sheets, financial statements, and complex tables from scanned or digital documents — with high accuracy, structure, and intelligence.

🔍 What Makes PiXTable Different?

PiXTable combines advanced OCR, computer vision, and Large Language Models (LLMs) to understand not just text — but financial meaning.

It can:
• Detect and extract tables from PDFs, images, and scans
• Understand rows, columns, merged cells, and totals
• Normalize financial data (assets, liabilities, equity)
• Convert unstructured tables into clean JSON, Excel, or ERP-ready formats

🧠 Powered by LLM Intelligence

Unlike traditional OCR, PiXTable uses LLMs trained on financial documents to:
• Identify balance sheet structures
• Classify financial fields automatically
• Detect inconsistencies and missing values
• Understand context (currency, period, fiscal year)

This means smarter extraction, not just text recognition.

⚙️ PiXTable API – Built for Developers & Enterprises

PiXTable is available as a secure, scalable API:
• REST / JSON interface
• Easy integration with banking, ERP, BI, and risk systems
• Cloud, on-prem, or hybrid deployment
• GDPR & enterprise-grade security

🏦 Use Cases

• Banks & Financial Institutions – automated balance sheet analysis
• Fintech – credit scoring & risk assessment
• Accounting & Audit firms – faster financial reporting
• Enterprises – ERP automation & financial consolidation

🚀 Why PiXTable?

Because financial data deserves accuracy, speed, and intelligence.

PiXTable transforms balance sheets from static documents into actionable, structured financial intelligence.

🔗 Powered by PixOCR | Built by Pixemantic

If you want to see PiXTable in action, try our demo or integrate the API today.

Pixemantic – Turning documents into decisions.

📩 Contact: [email protected]
🌐 Website: www.pixemantic.com

Address

Tunis
1053

Telephone

+21695610011

Website

http://pixemantic.com/

Alerts

Be the first to know and let us send you an email when Pixemantic posts news and promotions. Your email address will not be used for any other purpose, and you can unsubscribe at any time.

Contact The Business

Send a message to Pixemantic:

Pixemantic