Associate Consultant, AI Platforms at Sia Partners

Cédric Brzyski

Applied AI Engineer

EPITA engineering graduate in Artificial Intelligence and Data Science, now an Associate Consultant in AI Platforms within Sia Partners’ GenAI Solutions practice. I design and build AI applications and backend systems from business need to production.

Download résumé
LocationSaint-Maur-des-Fossés, France
Emailcbrzyski2@gmail.com
LanguagesFrench (native), English (C1)
01

Profile

My projects combine applied machine learning with software engineering: designing useful LLM workflows, working with visual and sensor data, and building interfaces that make technical systems usable. I am particularly interested in practical AI systems that operate reliably in constrained or professional environments.

Technical competencies

AI & LLMsAgentic AI, RAG, LLM orchestration, LangChain, Hugging Face, vector databases
Machine learningPyTorch, TensorFlow, scikit-learn, deep learning, computer vision, reinforcement learning
Software engineeringPython, C++, C, C#, SQL, JavaScript, TypeScript, Kotlin
Platform & deliveryFastAPI, Docker, AWS, Linux, Git, CI/CD, REST APIs, MLflow, ROS2
02

Experiences

Professional experiences

Sia Partners · Paris

Associate Consultant, AI Platforms

Part of the GenAI Solutions practice within the AI Platforms & Data Science team, contributing to enterprise AI platform strategy, design and delivery.

  • Translate client needs into robust AI products, automated workflows and enterprise integrations that connect business requirements with implementation.
  • Contribute across the GenAI platform lifecycle, from opportunity framing and solution design to delivery and adoption support.
  • Use LLM platforms and cloud infrastructure to build scalable, production-oriented AI solutions for clients across industries.
GenAI platforms · LLMs · Python · Cloud · Consulting methods

Safran · Paris

AI Engineer Intern

Translated ambiguous business requirements into production-ready AI workflows for Internal Audit teams.

  • Designed agentic applications for document analysis, findings extraction and automated reporting with Python, FastAPI, LangChain and AWS Bedrock.
  • Developed and deployed full-stack backend services on AWS and OpenShift ROSA with Terraform and Docker, within enterprise security and infrastructure constraints.
  • Worked directly with end users through discovery, development and deployment to refine requirements and workflows.
Python · FastAPI · LangChain · AWS Bedrock · AWS · OpenShift ROSA · Terraform · Docker

Inserm · Paris

R&D Mixed Reality & Software Intern

Developed real-time VR and AR interfaces for medical simulation.

  • Connected physical telemetry with interactive digital environments.
  • Optimized low-latency communication and telemetry-processing pipelines.
  • Stabilized real-time processing of multidimensional data for simulation workflows.
VR/AR · Real-time telemetry · Network optimization

Systhen · Fontenay-sous-Bois

Software & Operations Intern

Supported the optimization of internal administrative processes.

  • Worked with the Odoo ERP environment.
  • Created standardized internal documentation systems.
Odoo · Process documentation

Signaturit France · Paris

Partnership Intern

Contributed to reporting and partnership operations.

  • Built automated Power BI dashboards.
  • Contributed to strategic commercial and partnership meetings.
Power BI · Business intelligence
03

Projects

Selected technical work

Pinned repository

flightrec

Harness-agnostic flight recorder for AI coding agents. It captures model calls, file edits and shell commands in one replayable timeline, with the ability to fork a session from any recorded step.

Pinned repository

FinResearchAgents

Multi-agent equity-research system grounded in SEC filings: hybrid RAG, deterministic finance tools and an independent critic produce sourced memos with traceable numerical claims.

Pinned repository

Mistral Vibe

2nd-place overall project at the Mistral AI Hackathon. Extended Mistral’s Vibe CLI with reusable agent skills, browser automation and intelligent routing between local and cloud models.

Pinned repository

OfflineLingo

Offline Android speech-to-text translation application for emergency responders, using on-device speech recognition and local language models for private, low-latency multilingual communication.

Pinned repository

Autonomous Manipulation

Robotic manipulation pipelines combining imitation learning, computer vision and perception algorithms for real-world tasks, developed for the AMD Open Robotics Hackathon.

Pinned repository

Ovarian Cancer Segmentation

Deep-learning semantic-segmentation project for medical images, developed as a PINKCC Challenge finalist among more than 40 teams.

04

Hackathons & challenges

Selected competitive and collaborative work

Gemma 4 Hackathon

Gemmory

Private on-device memory assistant built with Gemma 4 and LiteRT-LM, including knowledge graphs, streaming generation and Room persistence.

OutcomeLocal Android AI memory vaultSource code

Mistral AI Hackathon

Mistral Vibe

Extended Mistral’s Vibe CLI with reusable agent skills, browser automation and intelligent routing between local and cloud models.

Outcome2nd place overallSource code

EDTH Berlin Hackathon

OfflineLingo

Emergency translation application using whisper.cpp and llama.cpp with a quantized Qwen 2.5 model and confidence cues.

OutcomeOffline Android translationSource code

HEC Fintech Hackathon

Real-time fraud detection

Phishing-risk scoring pipeline, Mistral explanations, Qdrant similarity search and a Chrome extension for real-time payment blocking.

OutcomeFull-stack prototypeSource code

XRPL Hackathon

PayTheTalent

Decentralized payment platform for talent marketplaces, built around secure XRPL wallet and transaction flows.

OutcomeFunctional MVPSource code

AMD Open Robotics Hackathon

Autonomous Manipulation

Designed robotic manipulation pipelines combining imitation learning, computer vision and perception algorithms for real-world tasks.

OutcomeTeam projectSource code

PINKCC Challenge

Ovarian Cancer Segmentation

U-Net segmentation models for tumors and metastases, combined through ensemble voting.

OutcomeFinalist · 12th / 42 teamsSource code

EPITA Kaggle Competition

Ship Image Classification

Custom CNN under 30 layers, with augmentation, regularization and staged training.

OutcomeTop 3 · 117 teamsSource code

GotaGoHack

LaD

Accessible web reader designed to support note reading for people with dyslexia and attention disorders.

OutcomeJury’s Favorite AwardSource code
05

Education

Academic background

EPITA

M.Eng. in Computer Science · Artificial Intelligence & Data Science

UQAC, Canada

Exchange semester in mathematics and computer science

Lycée Marcelin Berthelot

Scientific Baccalaureate, honours

06

Certifications

Professional learning credentials

07 / Contact

cbrzyski2@gmail.com

For an opportunity or collaboration, please get in touch by email or LinkedIn.

Ask about me

Experience, projects & ideas

Free AI onlyAI unavailable

Continue with predefined answers, without AI.

Classic Q&A · no AI

Predefined answers from the portfolio. Always available, no AI calls.

AI answers use OpenRouter and a free model provider. Please keep personal or confidential information out of your questions. Chats are not saved by this site.