About
Dr. Tang Daniel works at the intersection of Software Engineering and Artificial Intelligence. He completed his PhD at the University of Luxembourg from October 15, 2021 to July 1, 2025 (3 years and 9 months) and received the Excellent PhD Thesis Award.
He is the first author of CodeAgent (EMNLP'24), one of the earliest multi-agent LLM frameworks for autonomous code review; and the creator of PaperGuru, the first commercial recursive self-improvement product for training frontier-grade, highly efficient large models and solving complex problems.
His work has been published across SE, NLP, and ML venues including ICSE, FSE, ASE, ICML, TOSEM, EMSE, ACL, NAACL, EMNLP, AAAI, and ACM HEALTH. He serves as a PC member and reviewer for ICML, NeurIPS, ICLR, ACL, ICSE, and MSR, and as a journal reviewer for TOSEM, TSE, and EMSE.
Research Interests
- Recursive Self-Improvement & Learning
- Self-Morphable Model Training
- Unified Pre-Training, Post-Training, and Inference for Continuous Model Self-Evolution at Inference-Time Cost
- Dynamic Endless Agent Memory System
- LLM-Based Agents for Software Engineering
- Code Generation & Program Synthesis
- Automatic Program Repair
- Software Security & LLM Security
- Patch Representation & Explanation
Products & Flagship Work
CodeAgent
Pioneering multi-agent LLM framework where autonomous communicative agents collaborate on code review — detecting vulnerabilities, verifying consistency, and suggesting revisions. Presented as an invited talk at SAP Global Research and as a keynote at NLPAICS (Lancaster, UK).
PaperGuru
The first commercial recursive self-improvement product for training frontier-grade, highly efficient large models. PaperGuru solves complex problems through ultra-long-horizon reasoning, iterative self-correction, progressive capability growth, and continuous self-improvement. It surpassed 2,000 users in July 2026.
Latest News
- 2026-07 Milestone PaperGuru surpassed 2,000 users
- 2026-06 Funded TRACER selected for funding under the FNR TREK/JUMP 2026-1 call — maximum FNR contribution of €116,000
- 2026-05 Accepted BOOSTAPR accepted at ICML'26 — execution-grounded RL with dual reward models for program repair (40.7% on SWE-bench Verified)
- 2026-04 Accepted RETHiNK accepted at ACM Transactions on Computing for Healthcare — simulating human reasoning via multi-LLM debate for cognitive reframing
- 2026-03 Accepted Three posters accepted at FSE'26 Poster track — ConstantGuard, WasmCert, SPECTRA
- 2026-03 Accepted Two papers accepted at FSE'26 Ideas, Visions and Reflections track — The Watermark Paradox, Who Wrote This Patch?
- 2026-02 Accepted One poster accepted at ICSE'26 Poster track — Exploring LLMs for Compliance Violation Detection
- 2025-10 Service Invited as reviewer for ICLR'26, NeurIPS'26, ICML'26, and TOSEM'26
- 2025-08 Accepted Learning to Represent Code Changes (first author) accepted at EMSE'25
- 2025-08 Accepted Two papers accepted at ASE'25 NIER track — RAML, Measuring LLM Stability
- 2025-07 Award Completed PhD at the University of Luxembourg in 3 years and 9 months and received the Excellent PhD Thesis Award
- 2025-07 Accepted Two papers accepted at ICSME'25 NIER track — ExpertCache (first author), MalLoc
- 2025-05 Accepted LLMDA (first author) accepted by TOSEM'25; SynFix (first author) accepted by ACL'25 Findings
- 2025-05 Accepted Just-in-Time Detection of Silent Security Patches (first author) accepted by TOSEM'25, presented at ICSE'26 Journal-first
2026 // 9 accepted, including 1 Journal First
2025 // 9 papers
2024 // 10 papers
2023 & Earlier // 7 papers
TOSEM Revision // 5 manuscripts
Preprints // 5 selected manuscripts
Grants & Awards
An IDE with visual interaction and persistent memory empowering non-expert users in software development.
Key Contributor (Submitted)Trace-based Runtime Analysis for Code Evolution Review. Selected for funding by the Luxembourg National Research Fund under the FNR TREK/JUMP 2026-1 call. Project start by October 31, 2026.
Key Contributor · FundedSelected academic offers and recognition
Host-institution invitation to apply for the national overseas young talent program.
Nominated for the Chinese Academy of Sciences talent program.
Assistant Professor offer from the TU9 member university in Germany.
Associate Professor offer.
Services & Talks
Conference PC / Reviewer
- ICML2025–2026
- NeurIPS2025–2026
- ICLR2025–2026
- ACL Rolling Review2022–present
- AAAI2026
- ACM MM2024–2026
- ICSE · MSR · NLPCC—
- SOSP Artifact Evaluation Committee Member2021
Journal Reviewer
- ACM TOSEM2025–2026
- IEEE TSE2025–2026
- EMSE2025–2026
- Neurocomputing—
- Automated Software Engineering—
Invited Talks & Keynotes
- Keynote on CodeAgent — NLPAICS Workshop, Lancaster, UK2024
- CodeAgent — SAP Global Research Group (invited talk)2024
- CodeAgent demonstration — SnT Partnership Day2024
Open-Source Artifacts
- CodeAgent — multi-agent code reviewEMNLP'24
- Patcherizer — patch representationICSE'24
- MultiSEM — security patch detectionICLR'24
- BugRMSys — review-driven bug findingEMSE'24
- PaperGuru-Benchmark — recursive self-improvement benchmark2026