Back to developers
All on GitHub
Developer profile
Ajay Pravin Mahale
@designer-coderajay
About Ajay Pravin Mahale
AI/ML Researcher & Engineer | Mechanistic Interpretability | LLM Safety | NLP | Azure ML & MLOps | PyTorch | TransformerLens | Open to Full Time Roles
Public repositories
Recently updated work from this developer
designer-coderajay
glassbox-mech
Open-source EU AI Act Annex IV documentation toolkit. Mechanistic interpretability + circuit discovery for transformers. One function call generates a structured, hash-chained evidence package.
designer-coderajay
enterprise-agentic-ai-platform
Production-grade multi-agent AI platform with LangGraph v0.3, MCP servers, hybrid RAG (Qdrant + LlamaIndex), FastAPI WebSocket streaming, and Next.js real-time UI. Deployed on AWS EKS via Terraform.
designer-coderajay
Circuit-level-interpretability-evidence-does-not-survive-defensible-analytic-variation
Circuit-level interpretability evidence does not survive defensible analytic variation. Pre-registered multiverse over 15,840 specifications; the derived EU AI Act Annex IV claim flips across 73.2% of specification pairs. arXiv:2608.13754
designer-coderajay
Portfolio
Portfolio
designer-coderajay
designer-coderajay
Config files for my GitHub profile.
designer-coderajay
Causal-Attribution-for-Agentic-Decisions
Post-hoc causal attribution over LLM-agent trajectories. Two total effects that both fail, a closed form for coupling once contexts diverge, and a traceability specification read against Regulation (EU) 2024/1689. Pre-registered experiment published in full and not run.