CV

Education

---

Ph.D in Computer Science	2019-2022
University of Toulouse III - Paul Sabatier	Toulouse, France
- Thesis Topic Deep Learning for Information retrieval
- Advisors Jose G.Moreno, Mohand Boughanem
- Area of study Deep Learning, Information Retrieval (IR), and Natural Language Processing (NLP)

Master's degree in Computer Science	2018-2019
Higher National School of Computer Science (ESI)	Algiers, Algeria
- Thesis Topic Real-Time Tweet Summarization
- Advisors Lynda Said Lhadj, Mohand Boughanem
- Area of study Deep Learning, Information Retrieval (IR), and Natural Language Processing (NLP)

Engineering degree in Computer Science	2014-2019
Higher National School of Computer Science (ESI)	Algiers, Algeria
Majored in Information Systems & Software

Work experience

---

Data Scientist, NLP Specialist	October 2023, present
Mindflow	Paris, France
Designing no-code solutions powered by Generative AI agents to streamline task orchestration in cybersecurity operations. Designed and deployed stateful multi-agent systems leveraging Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG), enabling seamless natural language interactions with over 3,000 API integrations. [Patent Pending] Evaluated different retrieval approaches for RAG experimenting with sparse, dense, and hybrid models to improve retrieval accuracy and reduce latency. Studied various multi-agent workflows, including collaboration through shared memory and hierarchical agent interactions to enhance scalability and reduce cost. Designed an evaluation framework for assessing the planning and tool-use capabilities of LLMs, ensuring high-quality performance and decision-making in complex scenarios.

PostDoc Researcher in NLP	December 2022, September 2023
LIG, University of Grenoble Alps	Grenoble, France
Enhancing the systematic compositional generalization capabilities of sequence-to-sequence (seq2seq) models by integrating structural (syntactic) information into the decoding process through hyperbolic representations of dependency trees.

Researcher in NLP/IR	June 2021, January 2022
Max Plank Institute (MPI) for Informatics	Saarbrücken, Germany
Studied advancements with the ColBERT architecture, which relies on token-level representations with late interactions for document ranking. Proposed a novel approach for distilling ColBERT’s contextualized token embeddings into a more explicit and principled process by aggregating a finite set of frozen, pre-trained term-topic embeddings, with each term-topic capturing a contextual topic of a token.

Researcher in NLP/IR	December 2018, September 2019
Institut de Recherche en Informatique de Toulouse (IRIT)	Toulouse, France
Development of a tweet summarization approach based on Deep Learning Models. Manipulation of various state-of-the-art Models for both tweet representation and relevance estimation of tweets with respect to users' interests.

Skills

---

Coding Python, C, Typescript, JavaScript and PL/SQL

Libraries Pytorch, TensorFlow, Sickit-learn, Transformers, Sentence-Transformers, Faiss, fairseq

Operating Systems Microsoft Windows, Linux and other UNIX variants

Version Control Github

Agile Methodologies Scrum, Kanban

Languages English (native), French (native), Arabic (professional), Kabyle (native)

Publications

---

MarkedBERT: Integrating Traditional IR Cues in Pre-trained Language Models for Passage Retrieval

Highlighting exact matching via marking strategies for ad hoc document ranking with pretrained contextualized language models

Deep learning for information retrieval: studying relevant signals for ad hoc search based on transformer models

A Study of Term-Topic Embeddings for Ranking

Teaching

---

Data Structures and Fundamental Algorithms

Database Programming and administration

Hobbies

---

Photography | Reading | Drawing | Electronics

---

Download Resume