Hi, I'm Anurag Mishra

AI Engineer and researcher focused on mechanistic interpretability, LLM reliability, and high-performance ML systems.

Published Venues ICLR 2026 (Sole Author) IEEE (Co-authored)

I build and ship production AI systems, and I study how those systems work internally. My primary focus is mechanistic interpretability: tracing transformer circuits, testing causal hypotheses, and designing targeted interventions that improve behavior without losing efficiency.

I previously completed an M.S. in Artificial Intelligence at Rochester Institute of Technology (May 2025), where my work spanned multimodal deepfake detection, LLM evaluation and robustness, and performance-aware training workflows.

Download CV Download Resume

Research Focus

→ Mechanistic Interpretability and Causal Circuit Analysis

Activation tracing, attention-head profiling, and intervention-based analysis to identify task-relevant transformer mechanisms and measure causal contribution to final model behavior.

→ LLM Evaluation, Reliability, and Hallucination Reduction

Benchmark-driven evaluation across GPT-4, Claude, Llama, and Mistral families with stress testing, retrieval-augmented prompting, and regression checks for reasoning and factual consistency.

→ ML Systems Performance Engineering

Latency/throughput optimization, profiling, and regression debugging for production AI services, including model serving, asynchronous orchestration, and memory-aware optimization.

→ Multimodal Detection and Robustness

Multimodal CNN-transformer architectures for lip-sync manipulation detection, synthetic data generation, and robust out-of-distribution evaluation under realistic deployment constraints.

Professional and Research Experience

Industrial Experience

AI Engineer

Eaton Ventures (Rochester Appliances) • Aug 2025 - Present

Architected and deployed production AI services supporting 10K+ daily requests at 99.9% uptime.
Reduced end-to-end API latency by 35% through request-path and model-serving optimization.
Implemented CI/CD validation and regression pipelines that reduced release-cycle time by 60%.

Research Experience

Machine Learning Research Assistant, Mechanistic Interpretability

Rochester Institute of Technology • Jan 2025 - May 2025

Built activation tracing tooling in PyTorch/C++ for layer-wise transformer analysis.
Identified summarization-relevant circuits in layers 2, 3, and 5.
Applied targeted LoRA interventions with 40% faster convergence and 75% fewer trainable parameters.

Machine Learning Research Assistant, LLM Evaluation and Robustness

RIT Office of the Provost • Oct 2024 - Aug 2025

Designed evaluation harnesses for frontier models across reasoning, factuality, and tool calling.
Developed retrieval-augmented prompting workflows that reduced hallucination by 35%.
Built reproducible benchmark suites for quality and runtime regression tracking.

Machine Learning Research Assistant, DeFake Project

Rochester Institute of Technology • Oct 2024 - Aug 2025

Developed multimodal CNN + transformer systems on 18,000+ videos for lip-sync manipulation detection.
Implemented custom C++/CUDA tensor kernels that improved training efficiency by 40%.
Expanded stress-test data with 50K+ synthetic samples and improved OOD generalization by 18%.