Akshay maru

I’m a Senior AI Engineer with 7+ years building production AI systems in high-stakes domains. I founded and exited an AI SaaS with 40K users across 25 countries, and I specialize in RAG hardening, LLM security (OWASP LLM Top 10), and secure LLMOps pipeline design.

At JupiterOne, I’m architecting secure AI for enterprise security teams, where a hallucinated finding can trigger a false security incident. I’ve hardened a hybrid RAG pipeline against prompt injection and data exfiltration using OWASP-aligned semantic guardrails, reduced prompt token load ~45% with hierarchical token-aware summarization and adaptive chunk pruning, and built production evaluation pipelines to prevent silent model degradation.

Previously at Remote Leaps, I engineered a controlled-generation resume engine for ATS compatibility and a zero-hallucination production requirement, plus a fine-tuned resume scoring model using curriculum learning and hybrid job-matching (embeddings + BM25). I also scaled to 40K+ users across 25 countries, led a 4-person engineering team end-to-end, and contribute open-source projects that demonstrate verifiable LLM security and on-device AI patterns.