Graph based AST-tree analysis of arbitrary codebases for use in AI assistant coding pipelines - BC-965

Genre de projet: Innovation
Discipline(s) souhaitée(s): Génie - informatique / électrique, Génie, Informatique, Sciences mathématiques, Mathématiques
Entreprise: Farpoint Technologies Inc.
Durée du projet: 6 mois à 1 an
Date souhaitée de début: Dès que possible
Langue exigée: Anglais
Emplacement(s): Vancouver, BC, Canada
Nombre de postes: 2
Niveau de scolarité désiré: MaîtriseDoctorat
Ouvert aux candidatures de personnes inscrites à un établissement à l’extérieur du Canada: No

Au sujet de l’entreprise: 

Farpoint Technologies is a leading AI digital transformation consulting company that empowers top-tier organizations to build the AI-assisted workforce of the future. We specialize in providing consulting services to large public, private, and government entities, helping them create AI-accelerated workflows and innovative solutions. Our expertise includes leveraging LLMs, diffusion models, multimodal models, and executing special projects.

Veuillez décrire le projet.: 

This project develops a multi-model AI architecture designed to handle distinct software engineering tasks—system-level planning, function-specific code generation, and SQL database optimization. Unlike single-model solutions, this innovative approach assigns specialized roles to multiple models, utilizing a large planner model for architecture and smaller specialist models for granular tasks. The result is a flexible, efficient, and highly performant AI-powered pipeline, capable of generating accurate code and optimized SQL schemas at lower computational costs and higher specificity.

Main tasks for the candidate:
• Define structured JSON schemas for clear communication between planner and executor models.
• Integrate a large (≥70B) planning model to produce system-level architectural outlines.
• Deploy smaller specialized models for detailed code generation and SQL optimization.
• Implement a lightweight task-routing system assigning roles based on model specialization.
• Perform comprehensive performance evaluation on real-world PHP/TypeScript repositories.

Methodology / techniques:
• Task routing and role assignment using structured JSON schemas.
• Prompt engineering tailored to model capabilities and tasks.
• Micro-service architectures using Python or Go and gRPC communications.
• Automated Abstract Syntax Tree (AST) analysis and code generation.
• SQL schema optimization techniques and query performance analysis.
• Evaluation metrics including RepoBench for accuracy, latency, and SQL performance.

Expertise ou compétences exigées: 

Required expertise/skills:
• Strong skills in prompt engineering and interaction design for multiple language models (e.g., GPT variants, LLaMA derivatives).
• Experience designing structured JSON schemas for inter-model communication.
• Proficiency with Python or Go micro-service architectures (gRPC communication, Docker Compose deployment).
• Familiarity with AST parsing and manipulation (tree-sitter, Babel, or similar).
• Fundamental understanding of SQL optimization techniques (indexing, normalization, query analysis).
• Experience conducting structured benchmarking (RepoBench, pass@k metrics, latency measurements).

Assets (optional):
• Knowledge of reinforcement learning basics for potential task-routing optimization.
• Experience with version control and CI/CD (e.g., GitHub Actions).