Arnav Raj

PEBS: Per-rater Empirical-Bayes Shrinkage for RLHF Reward-Model Calibration

Solo-authored. Introduced a lightweight statistical layer that calibrates RLHF reward models to each individual human annotator rather than collapsing thousands of raters into a single average. Improves alignment to genuine human preferences on standard pluralistic-alignment benchmarks while leaving the underlying reward model unchanged.

Retroactive Advantage Correction: Closed-Form V-Trace Bias Correction for Delay-Aware RLHF

Solo-authored. Proposed a correction primitive for production RLHF pipelines where reward signals arrive late, such as slow code verifiers, large judge ensembles, and queued human review. Allows policy training to continue using delayed feedback instead of discarding it, with a closed-form proof of unbiasedness and substantial improvement over standard wait-or-drop strategies.

KG-MuLQA: Multi-hop Question Answering over Knowledge Graphs for Long-Context Evaluation

Benchmark for evaluating multi-hop reasoning in long-context language models, built from knowledge graphs over real-world financial documents. Designed and implemented the end-to-end pipeline for question generation, answer synthesis, and evaluation across frontier LLMs.

Hyperbolic Geometry of Reasoning: Probing LLM Hidden States

Solo-authored. Interpretability study of how chain-of-thought reasoning models internally encode hierarchical structure. Showed that hyperbolic geometric probes recover this structure across the entire network while standard Euclidean probes fail in the deepest layers, evidence that reasoning models compress hierarchy into the representations responsible for the final answer.

LLM Code Agent Evaluation Framework

Comprehensive evaluation suite for LLM code generation inspired by SWE-bench and MLE-bench, achieving 73% task completion with iterative self-correction.

RL Agent for Code Optimization

PPO-based reinforcement learning agent that optimizes code performance through iterative refinement, achieving 35% runtime reduction and 20% memory improvement.

Hangman AI: Transformer-Based Game Solver

Transformer-driven Hangman solver achieving >60% success using character-level modeling, morphological augmentation, and multi-strategy guess selection.

Graph Neural Network for User Personality Prediction

Bipartite user–product interaction graph leveraged with GNN architecture to infer user personality traits with high accuracy.

Data Search & Retrieval Engine

Neural-augmented retrieval system combining classical inverted indexing with dense vector search for hybrid relevance scoring across semi-structured technical documents.

Context-Aware Spelling Correction

Noisy-channel + smoothed N-gram language modeling system achieving 88% accuracy on context-dependent spelling errors.

TotalRecall: Cognitive Health App

Cross-platform Flutter app with AI-assisted memory support workflows for Alzheimer’s patients, backed by Firebase services.

Advanced Analytic Tool

A modular analytical platform that ingests heterogeneous datasets (CSV/JSON/SQL streams) and provides extensible pipelines for preprocessing, feature engineering, and rapid experimentation with classical ML and lightweight deep models.

AI Player for Havannah Board Game

Monte Carlo Tree Search (MCTS) agent with UCB exploration achieving >80% win rate over strong RAVE-only baselines.

SDN-based Intelligent Network Controller

High‑throughput OpenFlow controller with proactive L2 learning, shortest‑path routing, and loop prevention for complex topologies.

Application-Layer Reliable Transport & Congestion Control

Implemented a reliable transport protocol over UDP plus Reno & CUBIC congestion control variants with detailed performance benchmarking.

OS Kernel Enhancements in xv6

Implemented a page swapping subsystem in the xv6 teaching OS, including victim selection, swap slot management, and page fault handling.

I'm a fourth-year dual degree student at IIT Delhi (B.Tech + M.Tech in Computer Science) working at the intersection of AI safety, LLM evaluation, and interpretability. My research focuses on understanding how language models reason and where they fail, particularly through geometric and probing-based approaches.

At Google DeepMind, I author supervised fine-tuning and evaluation data for internal Gemini models, designing rubrics that feed directly into the training-data quality pipeline. At Abundant AI, I build GPU-backed evaluation infrastructure for long-running research agents, and previously designed adversarial ML benchmarks that are now part of the training pipeline for three of the world's top five AI labs. My solo-authored paper on probing LLM hidden states with hyperbolic geometry was accepted at the ICLR 2026 Workshop (GRaM). Two further solo-authored papers, PEBS on per-rater calibration of RLHF reward models and Retroactive Advantage Correction on delay-aware RLHF, are under review at ICML 2026 workshops. I co-authored KG-MuLQA, a long-context multi-hop evaluation framework, accepted at ACL 2026.

Outside of research, I co-founded the AI Safety Club at IIT Delhi and completed alignment training through BlueDot Impact and ARENA. I enjoy working on problems where evaluation methodology, model behavior, and safety considerations intersect.

Download Resume (PDF) →