Research engineering for high-risk AI systems
Senior AI execution where technical uncertainty is expensive.
I help funded biotech, medical-AI, and frontier-AI teams turn difficult R&D into validated prototypes, production-oriented pipelines, and clear technical decisions.
Selected experience includes Meta Reality Labs, QuantumCyte, AGAT Software, patent-sensitive medical imaging work, spatial-AI research, LLM systems, and mathematics-driven research engineering.
Proof of work
Deep technical work across research, product, and diligence.
Meta Reality Labs
Spatial AI, AR/VR, multimodal embeddings, and wearable-device prototypes.
QuantumCyte
Medical-imaging AI, precision workflows, spatial alignment, and patent filings.
AGAT Software
AI R&D leadership, LLM systems, anomaly detection, and customer-facing AI capabilities.
Mathematics background
M.Sc. Mathematics, modeling, abstraction, research reasoning.
Best use case
When the hard part is not just shipping code.
I work where the risk is research uncertainty: messy scientific data, ambiguous model behavior, precision-sensitive imaging, spatial representation, or AI claims that need serious technical review.
- Research-heavy methods need practical validation
- Data is difficult, noisy, or non-standard
- AI output must connect to real workflows
- Senior judgment and hands-on implementation both matter
Engagements
Services
Medical Imaging / Computer Vision R&D
Primary focusFor: Biotech and medical-AI teams
Senior technical support for imaging-heavy products involving segmentation, registration, calibration, spatial alignment, microscopy, pathology, or precision image-processing workflows.
1–3 week sprint or monthly retained R&D
- Clear technical risk assessment
- Prototype or production-oriented imaging pipeline
- Improved spatial alignment, calibration, or workflow reliability
Research Engineering Sprint
For: AI startups and R&D teams
Focused research-to-product execution for teams that need a prototype, paper implementation, benchmark, feasibility answer, or architecture path through an ambiguous technical problem.
Usually scoped in 1–3 week blocks
- Feasibility assessment
- Working prototype or benchmark
- Technical constraints and failure modes
AI / CV Technical Due Diligence
For: Founders, investors, and technical leaders
Independent assessment of whether an AI, computer-vision, medical-imaging, or spatial-AI claim is technically real, scalable, defensible, and execution-ready.
Lightweight review, diligence memo, or deeper technical assessment
- Clear technical risk map
- Feasibility and scalability assessment
- Architecture and data-quality review
Selected work
Proof-backed case studies
Medical Imaging AI Pipeline for Precision Workflows
Medical Imaging AI
AI-based software connecting medical segmentation, scan interpretation, geometric calibration, spatial alignment, and precision output generation.
Close to 5 µm precision workflow /
2 patent filings
Spatial AI and Wearable Interaction Prototypes
Spatial AI / AR-VR
Research and prototype work involving wearable-device interaction, spatial representations, and multimodal embeddings for AR/VR environments.
Working demo device-oriented prototype /
NDA-sensitive research context
TrackEverything — Model-Agnostic Tracking Layer
Computer Vision / Tracking
A Python package that can wrap detection or classification models and improve video predictions using temporal consistency, tracking, and statistical evidence across frames.
Open-source technical package /
Model-agnostic integration design
Doppler Radar Target Classification
Signal Processing / Machine Learning
Signal-processing and machine-learning pipeline for Doppler-pulse radar target classification using spectrograms, filtering, and constrained deep-learning experimentation.
30th competition placement /
1,000+ participants
Fit filter
Serious technical problems only.
Best fit
- Biotech or medical-AI teams with difficult imaging, segmentation, registration, calibration, or spatial-alignment problems
- AI startups translating research ideas into working prototypes
- Founders, CTOs, or investors who need senior technical judgment before committing budget
- Teams that need hands-on implementation, not just strategy
Not a fit
- Commodity dashboard work
- Low-budget MVP factories
- Academic ghostwriting
- Crypto projects
- Vague AI automation work without a serious technical problem
How work starts
Engagement model
1. Technical assessment
Clarify feasibility, risks, data constraints, and execution path.
2. Sprint or architecture plan
Build a prototype, benchmark, or implementation roadmap.
3. Retainer or advisory
Provide ongoing senior R&D capacity and technical ownership.
Ready to de-risk the technical path?
Send a concise brief with the problem, current stack or data type, timeline, and what failure would cost.
Available for selected contract, advisory, and retained R&D engagements.
Request a technical assessment