Reza Shamji
Research Associate @ Zitnik Lab (Harvard Medical School) PhD applications due Fall 2026 (hopeful entry Fall 2027) — theoretical deep learning & alternatives to neural networks
◆ Looking for Collaborations
I’m actively seeking exposure to theoretical deep learning research before submitting PhD applications this fall. If you work on alternatives to gradient descent, transformer architecture theory, model implicit bias, and/or transfomer/SSM/alternative to transformer training dynamics—I’d love to connect. Early feedback and collaborative projects would help ground my research direction.
Research Vision
I’m driven by a fundamental question: Is gradient descent on deep neural networks the right path toward artificial intelligence?
My research interests span two interconnected directions:
1. LLM Reasoning & Knowledge Integration Currently, I study how large language models leverage external knowledge—particularly when knowledge helps vs. harms performance. I’m developing model-agnostic predictors to determine when retrieval-augmented systems should be deployed, understanding the inductive biases that govern when agent decision-making improves under external constraints.
2. Theoretical Deep Learning & Architecture Alternatives My longer-term vision is to rethink artificial intelligence from first principles. I believe understanding the mathematics of transformers—attention mechanisms, QKV matrices, MLP/hidden state dynamics—is essential groundwork. But I’m skeptical that gradient descent is the most computationally representative path to general intelligence. During my PhD, I plan to:
- Explore alternative learning paradigms beyond gradient descent, potentially informed by (but not limited to) neuroscience
- Question core assumptions about what makes a system “intelligent,” building toward methods that may fundamentally differ from current neural network approaches
Preprints & Manuscripts
-
Democratizing AI Scientists using ToolUniverse · arXiv 2025 Zitnik Lab — Harvard Medical School · Co-author, advised by Marinka Žitnik. Open ecosystem enabling agent access to 1600+ scientific tools (aiscientist.tools). Partnered with Anthropic’s Claude. ToolUniverse serves as an official research connector within Claude to power scientific discovery. Featured in Nature, Science, and DecodingBio’s BioByte.
-
Understanding the Design Space and Cross-Modality Transfer for Vision-Language Models · ICML submission under review Kempner Institute (Harvard) · Co-author, advised by Yilun Du & Sham Kakade. Systematically mapped VLM design choices across image tokenizers, fusion architectures (Joint-Decoder, Cross-Attention, Mixture-of-Transformers), and layer-freezing recipes on a Qwen3 backbone, evaluating 50+ controlled configurations. Introduced three synthetic cross-modality transfer datasets (SpatialMap/Grid/Ring) with matched image-text task pairs to isolate AI reasoning from perception. Manuscript/code available upon request.
Research & Engineering
Zitnik Lab — Research Associate (Sept 2025 – Present)
Foundational LLM Health Reasoning & Knowledge Integration
- Studying inductive biases in retrieval-augmented clinical decision making — developing model-agnostic predictors for when external knowledge (e.g., knowledge graphs) improves vs. degrades LLM performance
- Implemented and evaluated GRPO to study post-training reasoning methods for health reasoning tasks External Stakeholder Collaboration
- Coordinating deliverables with Gates Foundation to translate research findings into deployable AI products
Infrastructure & ToolUniverse Integration
- Architected a hybrid semantic-keyword datastore (SQLite FTS5 + FAISS) with CLI + agent APIs, enabling researchers to convert lab-specific documents into AI-searchable collections for private or public use
- Transformed FAIR DCAT-AP metadata into a weekly-refreshed database of ~300 EU public-health datasets (21 tools; disease surveillance, cancer registries, mortality) with structured filtering and automated link extraction
- Integrated AlphaFold (UniProt) and HHS MyHealthFinder APIs to ground agent reasoning in structural biology and clinical guidance
- Driving CZI Biohub partnership to integrate foundation models into ToolUniverse for molecular/protein discovery Core infrastructure and findings underpin the ToolUniverse preprint (“Democratizing AI Scientists using ToolUniverse”).
Kempner Institute — ML Research Engineer Intern (Multimodal AI) (Jun – Sept 2025)
Advised by Yilun Du & Sham Kakade.
Large-Scale Ablation Study & Key Findings
- Image tokenizers trained with text-aware objectives consistently outperform text-blind tokenizers
- Modality-separated fusion (Mixture-of-Transformers) with freezing recipes that preserve base LLM knowledge improves out-of-domain generalization
- Cross-modality transfer is limited without tightly aligned/structured representations; image→text transfer stronger than text→image
Infrastructure & Benchmarking
- Built end-to-end VQA benchmarking for ChartQA, RealWorldQA, MMT-Bench, MathVista, DocVQA, TextVQA: dataset factories, OCR/table serialization, collate functions, and custom scorers (numeric relative-error, ANLS, MCQ)
- Implemented configurable query-key normalization (LayerNorm, RMSNorm, custom) to stabilize OCR/vision token processing
- Integrated Qwen3-8B into cross-attention fusion with selective freezing and developed a YAML-driven LR scheduler registry (cosine warmup, custom schedulers)
- Ran multi-node training/eval (FSDP/DDP on H100s) using Slurm with W&B monitoring
Findings underpin the VLM architecture design space preprint (ICML 2026 submission under review, manuscript and code available upon request).
Selected Projects
ChainEnv RL Benchmarks (JAX)
A compact sandbox to probe exploration under sparse rewards in a tunable 1-D chain (pure, vectorized JAX; laptop-friendly).
Repo: github.com/rezashamji/jax-chainenv-benchmarks
- Compares PPO, PQN, DDPG, SAC with deterministic evaluation vs exploratory training.
- Cleanly separates exploration from optimization and contrasts on-policy vs off-policy credit flow.
- Shows that as difficulty rises, learning becomes exploration-limited (not optimizer-limited); SAC/PQN help when discovery is easy, harder settings need discrete critics or intrinsic bonuses/parallelism.
Technical Skills
- Languages: Python, C++, Java, OCaml, SQL
- ML / DL: PyTorch, JAX, HF Transformers & Tokenizers, TorchVision, WebDataset
- Systems & Scaling: DDP, FSDP, DTensor, Slurm, NCCL, Docker, Conda
- Data / Tooling: Pandas, NumPy, YAML, Weights & Biases
- Hardware: multi-node H100 / A100 (e.g., 4×4 GPUs with FSDP)
Education
Harvard University — A.B. in Computer Science (Secondary in Economics) May 2025
Relevant Coursework: Algorithms & Limitations, Distributed Systems & Machine Organization, Semantics of Programming Languages, Probability, Linear Algebra
Languages
English (native) · Mandarin (fluent)
Contact & Links
- Email: reza_shamji@hms.harvard.edu
- LinkedIn: LinkedIn Profile
- GitHub: GitHub Profile
- CV: Download CV (PDF)