Projects
Explore the AI systems, software products, research and technical experiments I've led, developed and contributed to - from system architecture and implementation to evaluation and real-world applications.
Architecture diagram
AI systems
A Hebrew conversational assistant, built by a three-student course team, that reads tax documents, extracts and checks their fields, and tells a salaried worker what is still missing for a refund claim. It is not tax advice.
Workflow diagram
AI systems
A 52-clip benchmark of when SAM 2 video segmentation fails: it copes with extreme changes of scale, but breaks when the object itself is cut, poured or melted.
Workflow diagram
AI systems
The DDPM algorithm written from its equations and trained to generate digits, then one projection added to a pretrained face model's sampler to inpaint, super-resolve and colourise without training.
Workflow diagram
Product engineering
A bilingual (Hebrew and English) aggregator of published Israeli election polls, with its averaging method and per-poll sources published openly.
Workflow diagram
Data science & ML
Optimal quoting for a market maker carrying inventory risk: two stochastic-control problems, each solved and cross-checked by independent methods.
Workflow diagram
Data science & ML
The two camera poses and the visible 3-D corners of an 'impossible' triangle, recovered from two synthetic renders with epipolar geometry implemented from scratch in NumPy.
Workflow diagram
Data science & ML
A super-resolution network trained at test time on nothing but the low-resolution image itself, measured honestly against bicubic upscaling.
Schematic, not data
Data at scale
A two-person course project: describe a stay in plain words, a language model turns the wish into a target profile, and nearest-neighbour search ranks about two million listings enriched with public safety, income, health and air-quality data.
Shorter write-ups of algorithms implemented from first principles, mostly in coursework: the method, my role, what was provided, and a result only where one was recorded.
Technical lab
Node classification on the Cora citation graph
Node classification on the Cora citation graph with a two-layer GraphSAGE model, written as the student part of a course framework.
Technical lab
Active learning on a labelling budget
An active-learning strategy for a fixed labelling budget: committee disagreement chooses what to label, and validated rebalancing and confidence-gated pseudo-labels help a minority class.
Technical lab
An MLP and a CNN without autograd, checked against PyTorch
A working MLP and CNN written without autograd, each checked gradient by gradient against PyTorch and trained to the same accuracy as PyTorch's own version.
Technical lab
Trading detector calls for speed in a person tracker
A person tracker that runs a detector only every K frames and follows boxes with optical flow in between, with the box update and the choice of K worked out in a course assignment.
