// Le Laboratoire Kalkin

Gauraang

AI Researcher & Technologist

PhD Researcher  ·  Multi-Agent LLM Systems  ·  AI Consulting

Explore Work Get in Touch

Founder, Le Laboratoire Kalkin

Master's graduate (McGill '25) and former market-research PM, currently pursuing a PhD in Building Engineering. Work focuses on prototyping multi-agent LLM systems for hallucination mitigation, building Building-engineering ensemble models, and exploring creative AI in music and astrology.

At Le Laboratoire Kalkin, the work spans applied machine learning, artificial intelligence, and software development across agriculture (Bioponix), legal tech, and fitness.

Full bio & CV

Services

AI Consulting

Strategy and architecture for teams adopting LLMs, RAG, and multi-agent systems. Privacy-first, on-premises when required.

  • LLM system design & architecture
  • RAG pipeline implementation
  • Hallucination mitigation strategy
  • AI readiness audits

Multi-Agent Builds

End-to-end development of multi-agent pipelines for research, automation, and data-driven decision-making.

  • Agent orchestration (SmolAgents, AutoGen)
  • Vector store integration (FAISS, Supabase)
  • On-prem LLM deployment
  • Financial forecasting pipelines

Coaching & Training

AI literacy workshops and hands-on coaching for individuals and teams. Practical, jargon-free, results-focused.

  • AI basics workshops
  • Prompt engineering training
  • Tool selection guidance
  • 1-on-1 technical coaching

Selected Work

GitHub · Research

Multi-Agent LLM System

Agent vector search using cosine & euclidean similarity with research triangulation to mitigate LLM hallucinations. On-prem deployment.

FAISS PCA RAG On-prem LLM
View on GitHub
GitHub · Systems

AIOS: Local AI Operating System

Modular on-premises AI OS combining ML and LLM pipelines for local inference without cloud dependency.

On-prem LLM ML Pipeline Python
View on GitHub
GitHub · Creative AI

Astrology with TinyLlama

Astrological prediction engine using real-time planetary movement data integrated with a fine-tuned TinyLlama model.

TinyLlama Planetary API LLM
View on GitHub
GitHub · Finance

Stock Market Ensemble with Sentiment Analysis

Ensemble forecasting combining technical indicators with NLP-based sentiment analysis of financial news.

Ensemble Learning Sentiment NLP LSTM
View on GitHub
Canva · Database Design

Recruitment Database

🎞PRESENTATION · CANVALoad Presentation
Database Design SQL ERD
Open in Canva
Canva · ML · Health

Dental Age Estimator

🎞PRESENTATION · CANVALoad Presentation
ML Evaluation Health Data Scikit-Learn
Open in Canva
Canva · Privacy · Research

Implicit Data Inference

🎞PRESENTATION · CANVALoad Presentation
Survey Study Privacy Signals Qualitative
Open in Canva

Speaking & Presentations

SPEAKER

Cracking the Formula for AI & ML

The AI Collective × McHacks & Plotly · McGill University · 2025

The AI hierarchy, pairing LLMs with supervised ML and RAG, and how vector embeddings expose the roots of hallucination. Delivered at McHacks (with Plotly) and McGill University.

View on LinkedIn →
SPEAKER

ML, LLMs, RAG & AI Agents

Simply Voting Inc. · Montréal · March 2026

Corporate workshop covering ML fundamentals, transformer architecture, RAG systems, and AI agent components, with live demos of automation tools.

View on LinkedIn →
ORGANIZER

The AI Collective × Deck Meetup

Montréal · April 2026 · 150+ attendees

AI community meetup co-organised with Deck, exploring trust in AI products and building confidence through incremental validation.

View on LinkedIn →
ORGANIZER

The AI Collective MTL

Montréal · The AI Collective

Community AI event organised by The AI Collective, bringing together Montréal's applied AI practitioners.

View on Luma →