Tarun Raheja

I work on understanding how neural networks learn and reason: the geometry of memory in RNNs, multi-step reasoning in LLMs, learning in physical systems. Lately that means LLM security, agent simulations, in-context learning for science, and sampling-based planning for robotics.

• Jul 2026Paper on complementary LLM defenses accepted to GameSec 2026
• Jul 2026Paper on information diffusion in LLM-agent simulations accepted to the COLM 2026 Workshop on Social Simulation with LLMs
• Jul 2026DimPFN accepted to the COLM 2026 Workshop on Language Models for Scientific Discovery
• Jul 2026Paper on open problems in sampling-based planning and control at the RSS 2026 SBO Workshop
• Jun 2026Invited to mentor startups at Conquest, BITS Pilani, Asia's largest non-diluting startup accelerator
• Mar 2026US Patent 12,572,536 (Cache-Generated FAQ Page) granted
• Feb 2026Abstract on curricula as compression accepted as a poster at EHBEA 2026
• Feb 2026Joined Accel as AI Resident to Shekhar Kirani, advising on AI investments and helping portfolio companies
• Jan 2026Preprint unifying RLHF and direct alignment methods posted to arXiv
• Jan 20251st place at the Exa AI Discovery Agents Hackathon (50+ teams)
• Jan 2025Alignment Trilemma accepted to the ICLR 2025 Workshop on Bidirectional Human-AI Alignment
• Dec 2024Paper on neural networks for option pricing accepted to the ICLR 2025 I Can't Believe It's Not Better Workshop
• Nov 20241st place at the Sentient Foundation Werewolf AGIThon (AGI House)
• Oct 2024Persistent homology for tactile recognition accepted as an oral at the NeurIPS 2024 Workshop on Touch Processing
• Oct 2024Survey on foundation models and continual learning accepted to the NeurIPS 2024 Workshop on Scalable Continual Learning
• May 2024Submitted MS thesis on learning in physical systems with Andrea Liu and Nachi Stern
• May 2024Paper on LLM reasoning via classification accepted to LREC-COLING 2024
• May 2024Joined Kipo AI as Founding ML Engineer (employee #2), later Head of ML
• May 2023Started ML research internship at Cisco Research
• Mar 2023QuTiP-BoFiN published in Physical Review Research
• Jan 2023Joined Andrea Liu's group at UPenn Physics
• Sep 2022Joined the Linguistic Data Consortium with Mark Liberman
• Aug 2022Started MS in Scientific Computing at UPenn
• Aug 2021Joined J.P. Morgan as Quantitative Researcher (Rates)
• Oct 2020QuTiP-BoFiN preprint posted to arXiv
• 2020Took part in research with NTT's Coherent Ising Machine group
• 2020CS50 Puzzle Day (Harvard × Facebook): global winner (8/8)
• 2020Joined tensoralpha GmbH as Quantitative Developer Intern
• 2019Invited to IBM's Qiskit Asia Camp in Japan
• Oct 2019Joined RIKEN Japan with Franco Nori
• Jul 2019Started research internship at Microsoft Research India
• May 2019Submitted undergraduate thesis on DMRG for Potts models with Sreejith GJ (IISER Pune)
• 2019WorldQuant Challenge: gold ranking for high-Sharpe alphas
• Jan 2019Started undergraduate thesis research at IISER Pune
• Dec 2018Offered a WorldQuant WebSim research consultant role for exceptional performance
• May 2018Selected for the ACM Summer School in Cryptology at ISI Calcutta
• May 2017Started R&D internship at Zero Labs (NLP)

• 
  Coherent Ising machines. Research with NTT's Coherent Ising Machine group on optical hardware for hard combinatorial optimization: each laser pulse in a fiber ring acts as a spin, and measurement-feedback couples the pulses until the light settles into a low-energy solution. NTT Research, 2020
• 
  Exceptional points in non-Hermitian quantum systems. Studied exceptional points and exotic properties of large non-Hermitian systems with an algorithm I developed. with Fabrizio Minganti, RIKEN
• 
  Noisy mean-field annealing. Tuned NMFA for NP-hard combinatorial problems using ideas from quantum thermodynamics. with Neill Lambert, Clemens Gneiting, Ken Funo, RIKEN
• 
  Quantum overlapping tomography. Built a tomography module using perfect hash families, based on Frank Wilczek's paper, and measured entanglement quality on IBM machines. with Rudy Raymond, IBM Research; Qiskit Asia Camp, Japan
• 
  PyFroid. A Pandas plugin that gets 100x speedups on some operations through lazy evaluation. with Karthik Ramachandra, MSR India
• 
  Speech event segmentation. Better segmentation metrics via quantile analysis of intersyllable durations, and point-process models of syllable events. with Neville Ryant, LDC
• 
  Ontology-aware medical concept extraction. BiLSTM-CRF and transformers with UMLS embeddings for EHR entity recognition. with Lyle Ungar, Penn
• 
  Neural network potentials. ResNet-based potentials for copper thermal expansion, validated in LAMMPS. with Paris Perdikaris, Penn
• 
  Quantum and classical key exchange protocols. Cryptology research at the ACM Summer School, one of the top 50 undergrads selected nationwide. ISI Calcutta
• 
  Computational geometry library. C++ implementations of convex hulls, polygon triangulation and point visibility. with Tathagata Ray, BITS Pilani
• 
  MPI parallel code optimization. PETSc-based aerofoil sensitivity analysis on hybrid GPU-CPU systems. with N. Anil, BITS Pilani
• 
  Data envelopment analysis for student performance. DEA-based efficiency analysis of student input factors. with Srinivasa Raju, BITS Pilani

• 2024Mastering the Art of Prompting for LLMs: Reducing Hallucination and Improving Reasoning (Cisco Outshift blog)
• Dec 2024Oral: persistent homology for tactile object recognition, NeurIPS 2024 Workshop on Touch Processing

Reviewer for ICLR (2025, 2026, 2027) and NeurIPS 2026, plus a dozen workshops.

I've been lucky to work with some truly amazing people:

Favorite read: HPMOR.