AI Tinkerers Hackathon Results - Hackathon: Benchmarking Small Language Models in the Real World
AI Tinkerers - Paris
Sign in

Hackathon: Benchmarking Small Language Models in the Real World

Sign in
Engineering Lead at Stealth AI Startup - Stealth AI Startup
Own AI-native architecture and backend pipelines remotely with no equity gimmicks.
All Jobs
Final round winners have been announced. View Results

Results

Final winners and finalists for Hackathon: Benchmarking Small Language Models in the Real World

Winners

Final Round
1st Place

Darkwood

Team consisting of a Sorbonne Executive MBA/CEO, a Polytech Paris-Saclay Tech Lead, and a C# game developer, skilled in Go, Symfony, React, and workflow automation.
“Excellent use of RAG and specialized Polars prompting with a 3B model, balancing high correctness (N) with efficiency (T/VRAM).” “The technical grounding in Polars syntax rules is outstanding.”
Watch Video
Mathieu
Victor-eliejah
Mirza
10x Data Scientist

Muon

Team led by a Senior ML Engineer and IMT Atlantique alumna with 6 years experience in RAG, MoEs, and deploying LLMs via GCP and BentoML.
“Muon excels with a clever ‘Pseudo-RAG’ routing system and robust post-processing, balancing high correctness with efficiency.” “The iterative few-shot curation and clean architecture make it a top contender.”
Imane
Best Startup

bluebull

Team consisting of a Ublo software engineer and 42 Paris alumna specializing in React, TypeScript, and Python-based AI agentic workflows.
“Excellent use of 4-bit quantization and a self-repair loop to maximize correctness while minimizing VRAM footprint...” “...perfectly aligning with the hackathon's efficiency-weighted scoring formula.”
Vasiliki

Finalists

Preliminary Round
Finalist

Polaris

Team consisting of an AI engineer, a Python RPA consultant, and a CY Tech/IIT Chicago graduate specializing in TypeScript, LLM orchestration, and SAP automation.
Finalist

bluebull

Team consisting of a Ublo software engineer and 42 Paris alumna specializing in React, TypeScript, and Python-based AI agentic workflows.
Finalist

Training Expert

Team consisting of senior data engineers from SNCF Réseau and PLF, skilled in Databricks, Spark, Python, and cloud infrastructure, building production-grade AI agent systems.
Finalist

Muon

Team led by a Senior ML Engineer and IMT Atlantique alumna with 6 years experience in RAG, MoEs, and deploying LLMs via GCP and BentoML.
Finalist

Darkwood

Team consisting of a Sorbonne Executive MBA/CEO, a Polytech Paris-Saclay Tech Lead, and a C# game developer, skilled in Go, Symfony, React, and workflow automation.

Prizes Offered

  • Replit Pro Subscription: Elevate your coding game!
  • 6 Months Alpic Business: Unlock enterprise-grade features including Instant Deploy, 2M Requests/Month, 1-Year Analytics Retention, SLA & Slack Connect, Custom Domains, and OAuth DCR Proxy (a $300/mo value!).
  • €2000 in Mistral AI credits to be shared among the winning team.

Special Category Awards:

  • Best Startup gets access to Cloudfare credits!
  • GitHub swag for the best solo 10x data scientist, celebrating the ultimate solo hacker, team of one, power of ten.
    Prepare to be rewarded for your ingenuity and hard work! ✨