Four papers at IEEE WF-IoT
Four presentations on October 26: CBOR-LD-Ex, IMU-JEPA, Compliance Algebra, and 1-bit LLM routers.
Research dossier / Melbourne, Florida
Building AI systems that know what they know.
I’m Muntaser Syed, a Lead Gen AI Engineer, researcher, and open-source builder. I work on trustworthy AI, agent infrastructure, edge intelligence, semantic data, and efficient machine learning.
44Scholarly outputs
25Featured PyPI packages
3Package ecosystems
2Devpost portfolios
Coming up
On the calendar.
Scheduled sessions, accepted research, and a call for papers.
Energy attribution at CloudCom
I’m presenting our containerized ML energy-attribution paper in CloudCom’s sustainability, efficiency, and security session, which runs 14:00–15:30 Paris time.
Four papers at IEEE ICTAI
I’m presenting Hashrope, Pollard, StegaQR, and our paper on privacy-preserving similarity search at IEEE ICTAI 2026.
Veracity-aware data exchange
Our Application Track paper on grounding JSON-LD in Subjective Logic is in the 10:00–12:00 Intelligent Data Systems session in Milpitas, California.
CryptEpi and CDXF
CryptEpi is accepted at ICDM’s TML4DM workshop during the November 12–15 conference. CDXF is accepted at ComComAp, December 14–17 in Paphos.
The water and carbon cost of sovereign AI
I’m confirmed to speak at the 2026 IEEE Smart City USA Conference about how siting and cooling choices shape AI infrastructure’s environmental impact.
Embodied AI, robotics, and IoT
I’m co-chairing the special track at FLAIRS-40, May 24–27, 2027. Register abstracts by January 22 and submit papers by February 1, both at 23:59 Anywhere on Earth.
Latest / October 2026
Recent work and recognition.
The Power of No
Presented at AiSalon Miami on October 8 at Mana Tech Loft 10: how Pollard lets AI agents refuse, pause, and prove what happened before real-world actions.
geniekit and Pollard 1.6.2
geniekit is a new Python package for harmonizing GENIE BPC clinicogenomic cohorts. Pollard 1.6.2 adds an optional governed Codex CLI adapter with replay support.
IEEE GHTC Reviewers Choice Award
Our Environmental Cost of Digital Sovereignty paper received the Reviewers Choice Award in the Research on the Broader Impacts of Engineering Efforts track.
Research map / 01
Research areas.
My work covers epistemic trust, compact data representation, edge computation, and practical AI infrastructure. The links below connect each area with its papers and software.
01
Trustworthy & semantic AI
Machine-readable confidence, provenance, temporal validity, compliance, and knowledge verification.
02Agent systems
Governed execution, context budgets, deliberative simulation, and reproducible agent infrastructure.
03Edge AI
Intelligence that fits constrained devices, networks, and real-world sensor systems.
04Efficient ML
Physics-grounded energy measurement, compact architectures, and frugal model deployment.
05Health & embodied intelligence
Wearable sensing, medical imaging, speech systems, robotics, and privacy-aware health AI.
06Data systems
Reasoning-carrying binary linked data, hash-augmented structures, and compact interchange for high-integrity machine knowledge.
Selected work / 02
Recent publications.
Recent publications and preprints, with links to papers, code, and presentation materials.
2026
Preprint · Oct 12
2026
Preprint · Oct 12
2026
Preprint · Aug 25
2026
Accepted · IEEE MLSP
2026
Accepted · IEEE ICMLA
2026
Published · IEEE IRI
Open source / 03
Major open-source projects.
The software page groups related packages and ports. It documents the problem, the implementation, and the available Python, Rust, or TypeScript release.
Chronofy
Temporal embeddings, decay-weighted retrieval, and knowledge-validity checks for time-aware RAG.
Tokenmaster + ctxmaster
Context-budget metering, prediction, visualization, compaction, and handoff decisions for LLM applications.
PythonRustnpm
Hashrope
Balanced hash-augmented ropes for sub-linear comparison, routing, repetition encoding, and bioinformatics.
Recognition / 04
Recognition and public profiles.
Named one of Major League Hacking’s Top 50 hackers.MLH Top 50, Class of 2020
Collaborate / 05
Contact.
Email me about research collaboration, open-source work, technical talks, or applied AI projects.