A Transferable Variant Signal That Grocery Retrieval Cannot Reach
Accepted for an oral presentation · Osaka, Japan · December 2026
Artificial Intelligence · Ph.D. student
Emmanuel G. Maminta.
University of the Philippines Diliman
Metro Manila, Philippines
I am a Ph.D. student in Artificial Intelligence at the Ubiquitous Computing Laboratory, advised by Prof. Rowel O. Atienza. I currently work as an AI Developer at OpenText Philippines.
My research focuses on embodied AI, multimodal AI, and vision-language models (VLMs), with a strong emphasis on retail applications (RetailVision). My broader interests include hand–object interaction, egocentric and exocentric computer vision, robotics, and world models.
Updates
Recent news
-
2026
Joined the Cohere Labs Open Science Community, where my focus will be on multimodal AI and world models.
-
2026
Our paper, A Transferable Variant Signal That Grocery Retrieval Cannot Reach, has been accepted for an oral presentation at ACCV 2026 in Osaka, Japan, this December.
Oral presentation -
Sep 2026
Presented Late Interaction Ranking for Effective Closed-Catalog Grocery Product Retrieval at ECCV 2026 in Malmö, Sweden.
-
Aug 2026
Presented What Matters For Grocery Product Retrieval With Open Source Vision Language Models at ICPR 2026 in Lyon, France. The work also earned the RRPR reproducibility badge.
-
2025
Earned the NVIDIA-Certified Professional: Agentic AI and NVIDIA-Certified Associate: Generative AI LLMs certifications.
Earlier updates
- Nov 2025
Our Sari Sandbox work was featured in Voxel51’s Best of ICCV 2025. Watch the recording.
- Oct 2025
Presented Sari Sandbox at the ICCV 2025 Workshop on RetailVision in Honolulu, Hawai‘i, USA.
- 2025
Contributed the AnyMaths adapter to GEPA for prompt optimization in mathematical word problems and LLM reasoning.
- 2025
Released GEPA-Lite, a lightweight implementation of GEPA under the MIT License.
- 2025
Our paper, Sari Sandbox: A Virtual Retail Store Environment for Embodied AI Agents, was accepted at the ICCV 2025 Workshop on RetailVision. Paper · Project
- 2025
Presented research findings in a technical workshop with Analog Devices Philippines. Project
- 2025
Completed the M.Eng. in Artificial Intelligence program at the University of the Philippines. Capstone project
- 2024
Presented research with Nikko Carlo A. Yabut and Samsung Research & Development Institute Philippines at the first National Electrical, Electronics, and Computer Engineering Conference (NEECECon).
- 2023
Joined UCL as a graduate researcher on a collaborative project with Samsung Research & Development Institute Philippines on open-vocabulary object detection in vision-language models.
- 2022
Co-taught EEE 111 (Introduction to Programming and Computation) and EEE 121 (Data Structures and Algorithms) with Prof. Jhoanna Rhodette I. Pedrasa at UP Diliman.
- 2022
Began M.Eng.–Ph.D. studies as part of the pioneering Turing Batch in the University of the Philippines AI graduate program.
Retail applications
Research
Multimodal AI & VLMs
Vision-language models and multimodal learning for fine-grained grocery product retrieval and open-vocabulary recognition in retail environments.
Embodied AI
Agents that perceive, reason, and act in retail environments, with interests in robotics, world models, and simulation.
Hand–object interaction
Hand–object interaction and visual understanding from egocentric and exocentric perspectives, especially in retail settings.
Applied work
My work focuses on parameter-efficient fine-tuning (PEFT), agentic AI, and harness engineering.
Additional interests: resource-efficient pre-training, reinforcement learning for LLMs, reward modeling, preference learning, AI safety and alignment, chain-of-thought monitoring, and prompt optimization.
Late Interaction Ranking for Effective Closed-Catalog Grocery Product Retrieval
Presented in Malmö, Sweden · September 2026
What Matters For Grocery Product Retrieval With Open Source Vision Language Models
Presented in Lyon, France · August 2026
Sari Sandbox: A Virtual Retail Store Environment for Embodied AI Agents
Presented in Honolulu, Hawai‘i, USA · October 2025
Also available
Scalable Early Detection of Mental Health Signals in Social Media Text
2025 · Preprint, not peer reviewed
Open source
Software
OpenMPR
I started OpenMPR to share open-source recipes for fine-grained grocery product retrieval.
stable-worldmodel
I contributed to stable-worldmodel, an open-source platform for reproducible world model research and evaluation.
GEPA-Lite
A lightweight implementation of Genetic-Pareto prompt optimization, released under the MIT License.
GEPA · AnyMaths
A custom adapter contributed to GEPA for optimizing prompts on mathematical word problems and LLM reasoning.
Background
Education &
experience
- Current
AI Developer
OpenText Philippines
I build enterprise artificial intelligence products, agentic systems, and LLMOps/MLOps infrastructure.
- Ongoing
Ph.D. in Artificial Intelligence
University of the Philippines Diliman · Expected 2027
- 2025
Master of Engineering in Artificial Intelligence
University of the Philippines Diliman
- 2022
B.S. in Electrical Engineering, with honors
University of the Philippines Diliman
My background spans research and development, software engineering, and project management. I have collaborated with Samsung Research & Development Institute Philippines on open-vocabulary object detection and co-taught introductory programming and data structures at UP Diliman.
Credentials
Certifications
Get in touch
Let’s talk research.
For research inquiries and collaboration, you can reach me by email.