AI & ML interests

Agentic AI, Harness Engineering, Agentic Memory, Deterministic Agents, LLM Fine-tuning

Organization Card

šŸ¤– Palindrome Research Labs

Agentic systems from first principles. We focus on the four pillars of production-grade autonomous AI, replacing LLM-driven hallucinations with predictable, reproducible systems.

šŸ”¬ Our Research Areas

  • āš™ļø Deterministic Agents Replacing LLM-driven control flow with fixed pipelines (decomposition, classification, dispatch) to guarantee identical outputs for identical inputs with audit-grade provenance.
  • šŸ› ļø Agentic Harnesses Developing configurable agent frameworks featuring externalized tool specifications and strict, path-based security boundaries. Tools are treated as data, not code.
  • 🧠 Advanced Memory Subsystems Building persistent memory that scales beyond the context window using effective retrieval architecture.
  • 🧬 Specially Trained Models Performing parameter-efficient fine-tuning of open-weight models on real, grounded execution traces to achieve reliable, production-quality tool calling at consumer-hardware costs.

🌐 palindromerl.com

models 0

None public yet

datasets 0

None public yet