# Laksh Advani

> Staff Applied Scientist at ServiceNow in Seattle. 8+ years in ML. Works on RL post-training (RLVR, GRPO, process reward models), LLM agent reliability and recovery, evaluation, and safety.

## Experience
- ServiceNow, Staff Applied Scientist (2024–present)
- Qualtrics, Applied Scientist 2 (2021–2024)
- Capital One Labs, Applied Scientist (2019–2021)
- Sandia National Labs, Graduate Research Associate (2018–2019)
- M.S. Computer Science (ML/AI), University of Colorado Boulder

## Skills
- Post-training: RLVR, GRPO, KTO, DPO, PPO, Process reward models, RLAIF, Synthetic data, LoRA/QLoRA
- Agents and evals: Multi-agent orchestration, Tool-use RL environments, Context engineering, RAG, LLM-as-a-judge, Eval harness design, MCP, LangGraph
- Infrastructure: PyTorch, TorchTitan, DeepSpeed, Distillation, Quantization, vLLM, Kubernetes, Argo, AWS

## Publications
- From Confident Closing to Silent Failure: Characterizing False Success in LLM Agents — ICML 2026 Workshop on Failure Modes in Agentic AI
- Trajectory Guard: A Lightweight, Sequence-Aware Model for Real-Time Anomaly Detection in Agentic AI — AAAI 2026 Workshop on Trusted Agents
- When Small Models Are Right for Wrong Reasons: Process Verification for Trustworthy Agents — AAAI 2026 Workshop on Trusted Agents
- Reading the Plan: Multi-Horizon Action Prediction from Frozen LLM Hidden States — NeurIPS 2026 Workshops — accepted at 5 of 5 workshops
- lakshadvani at SemEval-2026 Task 11: A Neuro-Symbolic Approach to Content-Independent Syllogistic Reasoning — SemEval 2026 — Top 3, Subtask 1
- Effective Proxy for Human Labeling: Ensemble Disagreement Scores in Large Language Models for Industrial NLP — GEM Workshop @ EMNLP 2023
- C1 at SemEval-2020 Task 9: SentiMix: Sentiment Analysis for Code-Mixed Social Media Text Using Feature Engineering — SemEval 2020 @ COLING

## Reviewing
- NeurIPS 2026 Workshop IAB
- NeurIPS 2026 Workshop Verify-Agents
- ICML 2026 Workshop FAGEN

## Links
- [Website](https://lakshadvani.github.io/)
- [Machine-readable profile (MCP-style tools)](https://lakshadvani.github.io/agent.json)
- [Google Scholar](https://scholar.google.com/citations?hl=en&user=EBLv9hAAAAAJ)
- [LinkedIn](https://www.linkedin.com/in/lakshadvani/)
- [GitHub](https://github.com/lakshadvani)
- Contact: laksh.edu@gmail.com